We study the distance between the two rightmost particles in branching Brownian motion. Derrida and the second author have shown that the long-time limit d₁₂ of this random variable can be expressed in terms of PDEs related to the Fisher–KPP equation. We use such a representation to determine the sharp asymptotics of as a → +∞. These tail asymptotics were previously known to 'exponential order;' we discover an algebraic correction to this behavior.

We present new efficient \textit{projection-free} algorithms for online convex optimization (OCO), where by projection-free we refer to algorithms that avoid computing orthogonal projections onto the feasible set, and instead relay on different and potentially much more efficient oracles. While most state-of-the-art projection-free algorithms are based on the \textit{follow-the-leader} framework, our algorithms are fundamentally different and are based on the \textit{online gradient descent} algorithm with a novel and efficient approach to computing so-called \textit{infeasible projections}. As a consequence, we obtain the first projection-free algorithms which naturally yield \textit{adaptive regret} guarantees, i.e., regret bounds that hold w.r.t. any sub-interval of the sequence. Concretely, when assuming the availability of a linear optimization oracle (LOO) for the feasible set, on a sequence of length $T$, our algorithms guarantee $O(T^{3/4})$ adaptive regret and $O(T^{3/4})$ adaptive expected regret, for the full-information and bandit settings, respectively, using only $O(T)$ calls to the LOO. These bounds match the current state-of-the-art regret bounds for LOO-based projection-free OCO, which are \textit{not adaptive}. We also consider a new natural setting in which the feasible set is accessible through a separation oracle. We present algorithms which, using overall $O(T)$ calls to the separation oracle, guarantee $O(\sqrt{T})$ adaptive regret and $O(T^{3/4})$ adaptive expected regret for the full-information and bandit settings, respectively.

We consider optimization problems in which the goal is find a $k$-dimensional subspace of $R^n$, $k<<n$, which minimizes a convex and smooth loss. Such problemsgeneralize the fundamental task of principal component analysis (PCA) to include robust and sparse counterparts, and logistic PCA for binary data, among others. While this problem is not convex it admits natural algorithms with very efficient iterations and memory requirements, which is highly desired in high-dimensional regimes however, arguing about their fast convergence to a global optimal solution is difficult. On the other hand, there exists a simple convex relaxation for which convergence to the global optimum is straightforward, however corresponding algorithms are not efficient when the dimension is very large. In this work we present a natural deterministic sufficient condition so that the optimal solution to the convex relaxation is unique and is also the optimal solution to the original nonconvex problem. Mainly, we prove that under this condition, a natural highly-efficient nonconvex gradient method, which we refer to as \textit{gradient orthogonal iteration}, when initialized with a "warm-start", converges linearly for the nonconvex problem. We also establish similar results for the nonconvex projected gradient method, and the Frank-Wolfe method when applied to the convex relaxation. We conclude with empirical evidence on synthetic data which demonstrate the appeal of our approach.

We consider a population of Bayesian agents who share a common prior over some finite state space and each agent is exposed to some information about the state. We ask which distributions over empirical distributions of posteriors beliefs in the population are feasible. We provide a necessary and sufficient condition for feasibility. We apply this result in several domains. First, we study the problem of maximizing the polarization of beliefs in a population. Second, we provide a characterization of the feasible agent-symmetric product distributions of posteriors. Finally, we study an instance of a private Bayesian persuasion problem and provide a clean formula for the sender's optimal value.

A perfect matching in a hypergraph is a set of edges that partition the set of vertices. We study the complexity of deciding the existence of a perfect matching in orderable and separable hypergraphs. We show that the class of orderable hypergraphs is strictly contained in the class of separable hypergraphs. Accordingly, we show that for each fixed k, deciding perfect matching for orderable k-hypergraphs is polynomial time doable, but for each fixed \(k\ge 3\), it is NP-complete for separable hypergraphs.

BACKGROUND On January 2, 2022, Israel began administering a fourth dose of BNT162b2 vaccine (Pfizer-BioNTech) to people aged over 60 years and at-risk populations, who had received a third dose of vaccine at least 4 months earlier. The effect of the fourth dose on confirmed coronavirus 2019 disease (Covid-19) and severe illness are still unclear. METHODS We extracted data for the Omicron-dominated period January 15 through January 27, 2022, from the Israeli Ministry of Health database regarding 1,138,681 persons aged over 60 years and eligible for the fourth dose. We compared the rate of confirmed Covid-19 and severe illness between those who had received a fourth dose at least 12 days earlier, those who had received only three doses, and those 3 to 7 days after receiving the fourth dose. We used Poisson regression after adjusting for possible confounding factors. RESULTS The rate of confirmed infection was lower in people 12 or more days after their fourth dose than among those who received only three doses and those 3 to 7 days after vaccination by factors of 2.0 (95% confidence interval [CI], 2.0 to 2.1) and 1.9 (95% CI, 1.8 to 2.0), respectively. The rate of severe illness was lower by factors of 4.3 (95% CI, 2.4 to 7.6) and 4.0 (95% CI, 2.2 to 7.5). CONCLUSIONS Rates of confirmed Covid-19 and severe illness were lower following a fourth dose compared to only three doses.

Thinking is complex. Over the years, several types of methods and paradigms have developed across the psychological, cognitive, and neural sciences to study such complexity. A rapidly growing multidisciplinary quantitative field of network science offers quantitative methods to represent complex systems as networks, or graphs, and study the network properties of these systems. While the application of network science to study the brain has greatly advanced our understanding of the brains structure and function, the application of these tools to study cognition has been done to a much lesser account. This topic is a collection of papers that discuss the fruitfulness of applying network science to study cognition across a wide scope of research areas from generalist accounts of memory and encoding, to individual differences, to communities, and finally to cultural and individual change.

The aim of this paper is to solve a multi-period supplier selection and inventory lot-sizing problem with multiple products in a serial supply chain. Compared to previous models proposed in the literature, our research incorporates a richer cost structure involving joint replenishment costs for raw material replenishment and production, and a more realistic description of the transportation costs represented as a vector of full-truck load costs for different size trucks. This problem can be displayed graphically as a time-expanded transshipment network defined by nodes and arcs that can be reached by feasible material flows. First, we propose an integrated mixed integer linear programming model that minimizes the cost over the entire supply chain for a given planning horizon. The model determines the optimal dynamic supplier selection, inventory lot-sizing, and production schedule simultaneously. Second, a sequential approach is proposed to solve the same problem. That is, a production schedule is determined first, and then a supplier selection and replenishment strategy is obtained according to that predetermined schedule. Sensitivity analysis comparing the two approaches is performed. Results show that, even though the integrated approach achieves the minimum cost, the sequential approach may be suitable for solving large-scale instances of the problem as it requires less information sharing and generates a near-optimal solution with shorter implementation time and computational effort.

Augmented Business Process Management Systems (ABPMSs) are an emerging class of process-aware information systems that draws upon trustworthy AI technology. An ABPMS enhances the execution of business processes with the aim of making these processes more adaptable, proactive, explainable, and context-sensitive. This manifesto presents a vision for ABPMSs and discusses research challenges that need to be surmounted to realize this vision. To this end, we define the concept of ABPMS, we outline the lifecycle of processes within an ABPMS, we discuss core characteristics of an ABPMS, and we derive a set of challenges to realize systems with these characteristics.

We consider a scheduling problem on identical machines, where the cost for each machine is the total size of jobs assigned to it, excluding its largest job. The objective is to minimize the cost of the schedule, which is the maximum cost over all machines. We study online algorithms with and without migration. We design an algorithm with competitive ratio 3 for the purely online variant and provide a lower bound of 2.26953 on the competitive ratio of any online algorithm. For two machines, we find tight bounds of 2 on the competitive ratio. Additionally, we design a polynomial time approximation scheme for the variant with migration. We also briefly discuss the same online problem on uniformly related machines.

Projection-free optimization via different variants of the Frank–Wolfe method has become one of the cornerstones of large scale optimization for machine learning and computational statistics. Numerous applications within these fields involve the minimization of functions with self-concordance like properties. Such generalized self-concordant functions do not necessarily feature a Lipschitz continuous gradient, nor are they strongly convex, making them a challenging class of functions for first-order methods. Indeed, in a number of applications, such as inverse covariance estimation or distance-weighted discrimination problems in binary classification, the loss is given by a generalized self-concordant function having potentially unbounded curvature. For such problems projection-free minimization methods have no theoretical convergence guarantee. This paper closes this apparent gap in the literature by developing provably convergent Frank–Wolfe algorithms with standard \(\mathcal {O}(1/k)\) convergence rate guarantees. Based on these new insights, we show how these sublinearly convergent methods can be accelerated to yield linearly convergent projection-free methods, by either relying on the availability of a local liner minimization oracle, or a suitable modification of the away-step Frank–Wolfe method.

The association between early improvement and subsequent change in cognition is unexamined in antidementia clinical trials. We aimed to examine the consequences of early-response to antidementia medication in Alzheimer's disease. Participant-level data were analyzed from five pivotal clinical trials of donepezil for Alzheimer's disease lasting up to 24 weeks (N = 1917). Early-response was based on Alzheimer's Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) change scores under minus four from baseline to week six, otherwise classified non-response; then subgrouped by donepezil or placebo. The primary analysis tested the group differences in ADAS-Cog change from baseline for the interval after week six up to 24, based on a three-level mixed-effects model repeated measures (MMRM) model. Four models of increasing complexity were tested, and the most parsimonious model was examined in the primary analysis. The remaining models were tested in sensitivity analysis. In the analytic sample, 32.09% (N = 396/1234) of donepezil and 24.01% (N = 164/683) of placebo participants were classified as early responders, and 67.91% donepezil (N = 838/1234), 75.99% (N = 519/683) placebo participants were not. MMRM identified a statistically significant (P < 0.05) responder group effect. Marginal means (MM) demonstrated more improvement for the early responders (donepezil: MM = −4.13, 95% CI = -5.93, −2.32; placebo MM = 1.81, 95% CI = −4.12, 0.50), compared to non-early responders (donepezil MM = 0.05, 95% CI = -1.40, 1.51; placebo MM = 2.59, 95% CI = 0.99, 4.19). Results replicated in sensitivity analysis. Our results inform clinicians regarding the extent and consequences of early improvement in cognitive functioning and potentially contribute to treatment monitoring and the design of clinical trials for Alzheimer's disease.

Psychometric network analysis is an alternative theoretically-driven analytic approach that has the potential to conceptualize cognitive impairment in Alzheimer's disease differently than was previously assumed and consequently detect unknown treatment effects. Based on individual participant data, extracted from three double-blind, randomized placebo-controlled clinical trials, psychometric networks were computed on observed Alzheimer's Disease Assessment Scale Cognitive Subscale scores at baseline (N=1,554) and on predicted change scores at 24 weeks of follow-up for participants who received donepezil (N=797) or placebo (N=484). A novel conceptualization of cognitive impairment in Alzheimer's disease was displayed through the baseline network, that had 90% (n=27) positive statistically significant (p<0.05) associations, and a most central aspect of ideational praxis. Following 24 weeks, treatment effects emerged via the differences between the change score networks. The donepezil network had more statistically significant (p<0.05) positive associations and a higher global strength (n=15; S=1.22; p=0.03), than the placebo network (n=8; S=0.57). This suggests that for those who were treated with donepezil compared with placebo, cognition is a more unified construct. The main aspects of change in cognitive impairment were comprehension of spoken language for the donepezil network and spoken language ability for the placebo network. Comprehension of spoken language apears to be most sensitive to psychopharmaceutical interventions and should therefore be closely monitored. Overall, our psychometric network analysis presents a new conceptualization of cognitive impairment in Alzheimer's disease, points to previously unknown treatment effects and highlights well-defined aspects of cognitive impairment  that may translate into future treatment targets.

The problem of a planar vehicle moving on a surface, such as aerial drones or small naval vessels, can be treated as a series of trajectory planning problems between way-points. While nominally the movement between each two fourth-dimensional points (positions and velocities) can be treated as a 1D projection of the movement on the vector connecting the two points, in the presence of arbitrary disturbance the full problem on a plane must be considered. The mixed minimum time–energy optimal solution is now dependent on the value and direction of the disturbance, which naturally affects the structure and completion of the movement task. In this work, we address the minimum time–energy problem of a movement on a 2D plane with quadratic drag, under norm state (velocity) and norm control (acceleration) constraints. The structure and properties of the optimal solution are found and analyzed. The Pontryagin’s maximum principle (PMP) with control and state constraints is utilized. Simulations supporting the results are provided and compared with those of the open-source academic optimal control solver Falcon.m.

Experts and workers in many domains make multiple similar but independent decisions in sequence. Often, the serial position of the case in the sequence influences the decision. Explanations for these serial position effects tend to focus on the role of decision makers’ fatigue, but these effects emerge also when fatigue is unlikely. Here, we suggest that serial position effects can emerge due to decision makers’ motivation to be or to appear consistent. For example, to avoid having inconsistencies revealed, decisions may become more favorable towards the side that is more likely to put a decision under scrutiny. We focus on the legal domain in which many high-stakes decisions are made in sequence and in which there are clear institutional processes of decision scrutiny. We analyze two field datasets: 386,109 US immigration judges’ decisions on asylum requests and 20,796 jury decisions in 18th century London criminal court. We distinguish between five mechanisms that can drive serial position effects and examine their predictions for these settings. We find that consistent with motivation-based explanations of serial position effects, but inconsistent with fatigue-based explanations, in both settings decisions become more lenient as a function of serial position, an effect that persists over breaks and contrasts previous findings concerning sequential parole hearings. We further find, as is predicted by motivational accounts, that the leniency effect is stronger among more experienced decision makers.

Waiting-for-service is a central, typically detrimental, factor in service experiences, and multiple delays will most likely amplify customers' poor impressions of a service. Yet multi-delay experiences are commonly assessed via macro measurements, e.g., overall waiting, as opposed to micro measurements that account for individual delays, e.g., maximal or most-recent delay. Furthermore, the COVID-19 pandemic has exacerbated the challenges of controlling micro measures -- physical distancing has turned short queues and waits into rigid constraints. Our paper, motivated by a health screening clinic, jointly considers macro and micro measurements.

To improve customers' waiting experience, the clinic implemented two information technologies: automated customer routing system (ACRS) and SMS-based customer paging system (SCPS). However, as our empirical study of the clinic revealed, these implementations had no significant impact for three main reasons. First, ACRS reduced system flexibility and thus caused unintended idleness of resources. Second, SCPS improved non-bottleneck activities. Third, the clinic automated its sub-optimal practice, namely station-level first-come-first-served policy, leading to long delays towards the end of a customer's route.

Nevertheless, these initiatives could facilitate enhanced operations and improve waiting experience. To this end, via a stylized queueing model and a data-driven simulation model of the clinic, we analyze delays in an open-shop service network. Our models reveal that a priority-based buffer strategy, which accounts for both system- and station-level characteristics, improves both macro- and micro-level measurements. In particular, when prioritizing according to shortest expected remaining processing time priority-base buffer policy, performance improves at both micro-level by 41.9% and macro-level by 14.9%.

The quality of learning generally improves with the scale and diversity of data. Companies and institutions can therefore benefit from building models over shared data. Many cloud and blockchain platforms, as well as government initiatives, are interested in providing this type of service. These cooperative efforts face a challenge, which we call ``exclusivity attacks''. A firm can share distorted data, so that it learns the best model fit, but is also able to mislead others. We study protocols for long-term interactions and their vulnerability to these attacks, in particular for regression and clustering tasks. We conclude that the choice of protocol, as well as the number of Sybil identities an attacker may control, is material to vulnerability.

Smartphone-based inertial and magnetic sensors can be the basis for pedestrian navigation, whenever external positioning signals are limited or unavailable. Such navigation solutions are typically accomplished by a practice known as pedestrian dead reckoning, wherein step length and heading angle are estimated to form the horizontal trajectory of the user. One of the main challenges in these methods is the unknown misalignment between user’s forward axis and device’s frame, which imposes great difficulty in finding the user’s heading. In this paper, the problem of estimating walking direction is addressed by a learning-based approach. Specifically, a novel deep network architecture is designed for extracting the motion vector in the device coordinates, using accelerometer measurements. It consists of temporal convolutions and multi-scale attention layers, and involves rotation-invariant property that is analytically derived. The network model is integrated with a geometric calculation, employing gravity and geomagnetic directions, to convert the motion vector into heading angle relative to north. Extensive experiments of natural walking activity were conducted by a single pedestrian, with smartphones placed freely in various pockets. These were used for training the proposed model, in a user-specific approach. Finally, the entire framework was evaluated on unseen data, comparing the deep network against a mechanistic baseline method. Using the proposed network, with merely 5500 parameters, the resulting heading errors had a median value of 9.8°, which was lower by 2.5° compared to the baseline method.

Participatory budgeting (PB) is a democratic process for allocating funds to projects based on the votes of members of the community. Different rules have been used to aggregate participants' votes. Past research has studied the trade-off between notions of social welfare and fairness in the multi-winner setting (a special case of participatory budgeting with identical project costs) by Lackner and Skowron (2020). But there is little understanding of this trade-off in the more general PB setting. This paper provides a theoretical and empirical study of the worst-case guarantees of several common rules to better understand the trade-off between social welfare, representation. We show that many of the guarantees from the multi-winner setting do not generalize to the PB setting, and that the introduction of costs leads to substantially worse guarantees, thereby exacerbating the welfare-representation trade-off. We extend our theoretical analysis to studying how the requirement of proportionality over voting rules affects this trade-off. We further study how the requirement of proportionality over voting rules effects the guarantees on social welfare and representation. We study the latter point also empirically, both on real and synthetic datasets. We show that variants of the recently suggested voting rule Rule-X (which satisfies proportionality) do very well in practice both with respect to social welfare and representation.

The growing literature on how people learn to make decisions based on experience focuses on two types of paradigms. In one paradigm, people are faced with a choice, and must retrospectively consult past experience of similar choices to decide what to do. In the other paradigm, people are faced with a choice, and then have the opportunity prospectively to gather new experiences that might help them make that choice. The current paper examines the joint impact of both retrospective and prospective experiences. Two experiments reveal strong interactions. In Study 1, repeated experience with new samples appears to reduce sensitivity to the average outcome in the samples and enhances underweighting of rare events. Study 2 shows that repeated experience with pre-choice samples can reverse the impact of the new information (and decrease the tendency to select the alternative that provides the best outcome in the new sample). The results suggest that prospectively gathering new samples can have two, potentially contrasting, influences on choice: the first focuses on the sample’s face value and selects the option with the higher value in the new sample; by contrast, the second treats the new sample as a cue to recall similar prior experiences, which in turn drive choice. The paper concludes with a discussion of the possibility that part of the descriptive value of prospect theory reflects the fact that it summarizes the joint impact of similar “face-or-cue” processes.

Makespan minimization on parallel identical machines is a classical and intensively studied problem in scheduling, and a classic example for online algorithm analysis with Graham's famous list scheduling algorithm dating back to the 1960s. In this problem, jobs arrive over a list and upon an arrival, the algorithm needs to assign the job to a machine. The goal is to minimize the makespan, that is, the maximum machine load. In this paper, we consider the variant with an additional cardinality constraint: The algorithm may assign at most $k$ jobs to each machine where $k$ is part of the input. While the offline (strongly NP-hard) variant of cardinality constrained scheduling is well understood and an EPTAS exists here, no non-trivial results are known for the online variant. We fill this gap by making a comprehensive study of various different online models. First, we show that there is a constant competitive algorithm for the problem and further, present a lower bound of $2$ on the competitive ratio of any online algorithm. Motivated by the lower bound, we consider a semi-online variant where upon arrival of a job of size $p$, we are allowed to migrate jobs of total size at most a constant times $p$. This constant is called the migration factor of the algorithm. Algorithms with small migration factors are a common approach to bridge the performance of online algorithms and offline algorithms. One can obtain algorithms with a constant migration factor by rounding the size of each incoming job and then applying an ordinal algorithm to the resulting rounded instance. With this in mind, we also consider the framework of ordinal algorithms and characterize the competitive ratio that can be achieved using the aforementioned approaches.

Integrity constraints such as functional dependencies (FD) and multi-valued dependencies (MVD) are fundamental in database schema design. Likewise, probabilistic conditional independences (CI) are crucial for reasoning about multivariate probability distributions. The implication problem studies whether a set of constraints (antecedents) implies another constraint (consequent), and has been investigated in both the database and the AI literature, under the assumption that all constraints hold exactly. However, many applications today consider constraints that hold only approximately. In this paper we define an approximate implication as a linear inequality between the degree of satisfaction of the antecedents and consequent, and we study the relaxation problem: when does an exact implication relax to an approximate implication? We use information theory to define the degree of satisfaction, and prove several results. First, we show that any implication from a set of data dependencies (MVDs+FDs) can be relaxed to a simple linear inequality with a factor at most quadratic in the number of variables; when the consequent is an FD, the factor can be reduced to 1. Second, we prove that there exists an implication between CIs that does not admit any relaxation; however, we prove that every implication between CIs relaxes "in the limit". Then, we show that the implication problem for differential constraints in market basket analysis also admits a relaxation with a factor equal to 1. Finally, we show how some of the results in the paper can be derived using the I-measure theory, which relates between information theoretic measures and set theory. Our results recover, and sometimes extend, previously known results about the implication problem: the implication of MVDs and FDs can be checked by considering only 2-tuple relations.

Effective indoor wayfinding in the event of an emergency is key to guaranteeing safe and timely evacuation. However, despite the increasing number of evacuation studies, only a limited number focus on the influence of architectural elements. Through a virtual reality experiment, we create a link between human factors in indoor emergency wayfinding and architectural design by exploring interior wall transparency, evacuation starting points, and architectural landmarks. Our findings show that wall transparency only impacts wayfinding when combined with visibility at route starting points, and that staircases and ramps are the most significant architectural landmarks in emergency wayfinding. These differ from previous studies that primarily emphasize overall visibility in a building. Additionally, in cases of more complex evacuation scenarios with low visibility conditions, wayfinding necessitates a greater number of architectural features. These findings enhance the understanding of architectural designs as complex, multi-leveled systems with numerous distinct features that evoke a set of structural relationships in emergency situations.

We present a method for matching a text sentence from a given corpus to a given video clip and vice versa. Traditionally video and text matching is done by learning a shared embedding space and the encoding of one modality is independent of the other. In this work, we encode the dataset data in a way that takes into account the query's relevant information. The power of the method is demonstrated to arise from pooling the interaction data between words and frames. Since the encoding of the video clip depends on the sentence compared to it, the representation needs to be recomputed for each potential match. To this end, we propose an efficient shallow neural network. Its training employs a hierarchical triplet loss that is extendable to paragraph/video matching. The method is simple, provide explainability, and achieves a state-of-the-art-results, for both sentence-clip and video-text by a sizable margin across five different datasets: ActivityNet, DiDeMo, YouCook2, MSR-VTT, and LSMDC. We also show that our conditioned representation can be transferred to video-guided machine translation, where we improved the current results on VATEX. Source code is available at https://github.com/AmeenAli/VideoMatch.

Several recent studies have explored the interaction effects between topics, systems, corpora, and components when measuring retrieval effectiveness. However, all of these previous studies assume that a topic or information need is represented by a single query. In reality, users routinely reformulate queries to satisfy an information need. In recent years, there has been renewed interest in the notion of “query variations” which are essentially multiple user formulations for an information need. Like many retrieval models, some queries are highly effective while others are not. This is often an artifact of the collection being searched which might be more or less sensitive to word choice. Users rarely have perfect knowledge about the underlying collection, and so finding queries that work is often a trial-and-error process. In this work, we explore the fundamental problem of system interaction effects between collections, ranking models, and queries. To answer this important question, we formalize the analysis using ANalysis Of VAriance (ANOVA) models to measure multiple components effects across collections and topics by nesting multiple query variations within each topic. Our findings show that query formulations have a comparable effect size of the topic factor itself, which is known to be the factor with the greatest effect size in prior ANOVA studies. Both topic and formulation have a substantially larger effect size than any other factor, including the ranking algorithms and, surprisingly, even query expansion. This finding reinforces the importance of further research in understanding the role of query rewriting in IR related tasks.

Intertemporal trade-offs between small, temporally proximate benefits and larger, delayed benefits are frequent in our everyday lives. Most research on intertemporal choices has been focused on “one-shot” descriptive and hypothetical scenarios which involve large scales of both money and time. Yet, in daily life, many small intertemporal decisions are made repeatedly, based on real previous experiences. In the current article, we investigate small repeated intertemporal choices with actual delays, focusing on three established intertemporal effects—common difference, magnitude, and delay duration. In a preliminary study, we compared small and large scales of rewards and delays in a descriptive hypothetical setting. In the small-scale problems, participants did not exhibit significant sensitivity to small changes in durations, leading to increased patience. Using the same small-scale problems, in Study 2 we employed a purely experiential environment and found a significant sensitivity to durations and reduced patience. To further examine the impact of description versus experience on patience, in Study 3, we added full description of the intertemporal trade-offs to the experiential setting of Study 2. Participants’ willingness to wait was relatively high in the first descriptive trial but decreased drastically after experiencing delays. The current results demonstrate similarities and differences between small, experience-based and large, description-based intertemporal choices. While the three known intertemporal effects were observed in all settings, we found a substantial gap between decisions made based on description and those made based on experience. Experiencing short delays seems to increase time sensitivity, leading to reduced willingness to wait. (PsycInfo Database Record (c) 2022 APA, all rights reserved)

Background

Knowledge is limited regarding the association between very-late onset schizophrenia and the risks of risks of death and dementia. We aimed examine the associations between very-late onset schizophrenia and the risks of death and dementia.

Method

A prospective Israeli cohort study (N=94,120) of persons without either dementia or schizophrenia (2002 to 2012), aged 60 to 90 in 2012, were followed-up for dementia or death (2013 to 2017). During follow-up schizophrenia was classified as present from the age of the first ICD-9 schizophrenia diagnosis, otherwise as absent. The association between schizophrenia and the competing risk of death and dementia was quantified with Hazard Ratios (HR) and their 95% confidence intervals (CI), fitting six competing risk analyses, unadjusted (Unadj.) and adjusted (Adj.) with increasing complexity. We tested the robustness of the results in nine sensitivity analyses.

Result

The group with schizophrenia had higher rates of death (n=61, 18.5% vs. n=7,028, 7.5%, respectively) and dementia (n=64, 19.5% vs. n=5,962, 6.4%, respectively). In the primary analysis, the group with compared to the group without schizophrenia had higher risks of death (Unadj. HR= 3.10, 95% CI = 2.36, 4.06, P <.001; Adj. HR= 2.89, 95% CI = 2.15, 3.89; P<.001) and dementia (Unadj. HR = 3.81, 95% CI = 2.90, 4.99, P <.001; Adj. HR= 2.67, 95% CI = 1.82, 3.91; P<.001). The results remained significant (P <.05) in all sensitivity analyses, including among persons without antipsychotic exposure.

Conclusion

Very-late onset schizophrenia was associated with increased death and dementia risks, possibly due to accelerated aging.

This work studies an experimental design problem where $x$'s are to be selected with the goal of estimating a function $m(x)$, which is observed with noise. A linear model is fitted to $m(x)$ but it is not assumed that the model is correctly specified. It follows that the quantity of interest is the best linear approximation of $m(x)$, which is denoted by $\ell(x)$. It is shown that in this framework the ordinary least squares estimator typically leads to an inconsistent estimation of $\ell(x)$, and rather weighted least squares should be considered. An asymptotic minimax criterion is formulated for this estimator, and a design that minimizes the criterion is constructed. An important feature of this problem is that the $x$'s should be random, rather than fixed. Otherwise, the minimax risk is infinite. It is shown that the optimal random minimax design is different from its deterministic counterpart, which was studied previously, and a simulation study indicates that it generally performs better when $m(x)$ is a quadratic or a cubic function. Another finding is that when the variance of the noise goes to infinity, the random and deterministic minimax designs coincide. The results are illustrated for polynomial regression models and different generalizations are presented.

In $\mathbb{Z}^d$ with $d\ge 5$, we consider the time constant $\rho_u$ associated to the chemical distance in random interlacements at low intensity $u \ll 1$. We prove an upper bound of order $u^{-1/2}$ and a lower bound of order $u^{-1/2+\varepsilon}$. The upper bound agrees with the conjectured scale in which $u^{1/2}\rho_u$ converges to a constant multiple of the Euclidean norm, as $u\to 0$. Along the proof, we obtain a local lower bound on the chemical distance between the boundaries of two concentric boxes, which might be of independent interest. For both upper and lower bounds, the paper employs probabilistic bounds holding as $u\to 0$; these bounds can be relevant in future studies of the low-intensity geometry.

A data ecosystem (DE) offers a keystone-player or alliance-driven infrastructure that enables the interaction of different stakeholders and the resolution of interoperability issues among shared data. However, despite years of research in data governance and management, trustability is still affected by the absence of transparent and traceable data-driven pipelines. In this work, we focus on requirements and challenges that DEs face when ensuring data transparency. Requirements are derived from the data and organizational management, as well as from broader legal and ethical considerations. We propose a novel knowledge-driven DE architecture, providing the pillars for satisfying the analyzed requirements. We illustrate the potential of our proposal in a real-world scenario. Last, we discuss and rate the potential of the proposed architecture in the fulfillmentof these requirements.

BACKGROUND After promising initial results from the administration of a third (booster) dose of the BNT162b2 messenger RNA vaccine (Pfizer–BioNTech) to persons 60 years of age or older, the booster campaign in Israel was gradually expanded to persons in younger age groups who had received a second dose at least 5 months earlier. METHODS We extracted data for the period from July 30 to October 10, 2021, from the Israel Ministry of Health database regarding 4,696,865 persons 16 years of age or older who had received two doses of BNT162b2 at least 5 months earlier. In the primary analysis, we compared the rates of confirmed coronavirus disease 2019 (Covid-19), severe illness, and death among those who had received a booster dose at least 12 days earlier (booster group) with the rates among those who had not received a booster (nonbooster group). In a secondary analysis, we compared the rates in the booster group with the rates among those who had received a booster 3 to 7 days earlier (early postbooster group). We used Poisson regression models to estimate rate ratios after adjusting for possible confounding factors. RESULTS The rate of confirmed infection was lower in the booster group than in the nonbooster group by a factor of approximately 10 (range across five age groups, 9.0 to 17.2) and was lower in the booster group than in the early postbooster group by a factor of 4.9 to 10.8. The adjusted rate difference ranged from 57.0 to 89.5 infections per 100,000 person-days in the primary analysis and from 34.4 to 38.3 in the secondary analysis. The rates of severe illness in the primary and secondary analyses were lower in the booster group by a factor of 17.9 (95% confidence interval [CI], 15.1 to 21.2) and 6.5 (95% CI, 5.1 to 8.2), respectively, among those 60 years of age or older and by a factor of 21.7 (95% CI, 10.6 to 44.2) and 3.7 (95% CI, 1.3 to 10.2) among those 40 to 59 years of age. The adjusted rate difference in the primary and secondary analyses was 5.4 and 1.9 cases of severe illness per 100,000 person-days among those 60 years of age or older and 0.6 and 0.1 among those 40 to 59 years of age. Among those 60 years of age or older, mortality was lower by a factor of 14.7 (95% CI, 10.0 to 21.4) in the primary analysis and 4.9 (95% CI, 3.1 to 7.9) in the secondary analysis. The adjusted rate difference in the primary and secondary analyses was 2.1 and 0.8 deaths per 100,000 person-days. CONCLUSIONS Across the age groups studied, rates of confirmed Covid-19 and severe illness were substantially lower among participants who received a booster dose of the BNT162b2 vaccine than among those who did not.

Israel began administering a BNT162b2 booster dose to restore protection following the waning of the 2-dose vaccine. Biological studies have shown that a fresh booster leads to increased antibody levels compared to a fresh 2-dose vaccine, which may suggest increased effectiveness. To compare the real-world effectiveness of a fresh booster dose with that of a fresh 2-dose vaccine, we conduct a quasi-experimental study that compares populations that were eligible to receive the vaccine at different times due to age cutoff policies. Our analysis shows that a fresh booster increases protection against confirmed infection by 3.7 (95% CI: 2.7 to 5.2) fold compared to a fresh 2-dose vaccine.

We study Dominant-Strategy Incentive-Compatible (DSIC) revenue-maximizing auctions ("optimal" auctions) for a single-item and correlated private values. We give tight bounds on the ratio of the revenue of the optimal Ex-Post Individually Rational (EPIR) auction and the revenue of the optimal Ex-Interim Individually Rational (EIIR) auction. This bound is expressed as a non-decreasing function of the expected social welfare of the underlying distribution. In particular, we show a class of distributions on which this ratio cannot be lower bounded by any positive number. Thus, the restriction to EPIR auctions, which has been the de-facto standard in the computer science literature on auctions with correlated values, may significantly reduce the revenue that can be possibly extracted, as the revenue extracted by an EPIR auction might be an arbitrarily small fraction of the revenue extracted by an EIIR auction.

Historically, psychedelic drugs are known to modulate cognitive flexibility, a central aspect of cognition permitting adaptation to changing environmental demands. Despite proof suggesting phenomenological similarities between artificially-induced and actual psychedelic altered perception, experimental evidence is still lacking about whether the former is also able to modulate cognitive flexibility. To address this, we measure participants’ cognitive flexibility through behavioral tasks after the exposure to virtual reality panoramic videos and their hallucinatory-like counterparts generated by the DeepDream algorithm. Results show that the estimated semantic network has a flexible structure when preceded by altered videos. Crucially, following the simulated psychedelic exposure, individuals also show an attenuated contribution of the automatic process and chaotic dynamics underlying the decision process. This suggests that simulated altered perceptual phenomenology enhances cognitive flexibility, presumably due to a reorganization in the cognitive dynamics that facilitates the exploration of uncommon decision strategies and inhibits automated choices.

We analyze boundedly rational updating in a repeated interaction network model with binary actions and binary states. Agents form beliefs according to discretized DeGroot updating and apply a decision rule that assigns a (mixed) action to each belief. We first show that under weak assumptions, random decision rules are sufficient to achieve agreement in finite time in any strongly connected network. Our main result establishes that naive learning can be achieved in any large strongly connected network. That is, if beliefs satisfy a high level of inertia, then there exist corresponding decision rules coinciding with probability overmatching such that the eventual agreement action matches the true state, with a probability converging to one as the network size goes to infinity.

The success of deep neural nets heavily relies on their ability to encode complex relations between their input and their output. While this property serves to fit the training data well, it also obscures the mechanism that drives prediction. This study aims to reveal hidden concepts by employing an intervention mechanism that shifts the predicted class based on discrete variational autoencoders. An explanatory model then visualizes the encoded information from any hidden layer and its corresponding intervened representation. By the assessment of differences between the original representation and the intervened representation, one can determine the concepts that can alter the class, hence providing interpretability. We demonstrate the effectiveness of our approach on CelebA, where we show various visualizations for bias in the data and suggest different interventions to reveal and change bias.

Education is central to the acquisition of knowledge, such as when children learn new concepts. It is unknown, however, whether educational differences impact not only what concepts children learn, but how those concepts come to be represented in semantic memory—a system that supports higher cognitive functions, such as creative thinking. Here we leverage computational network science tools to study hidden knowledge structures of 67 Swiss schoolchildren from two distinct educational backgrounds—Montessori and traditional, matched on socioeconomic factors and nonverbal intelligence—to examine how educational experience shape semantic memory and creative thinking. We find that children experiencing Montessori education show a more flexible semantic network structure (high connectivity/short paths between concepts, less modularity) alongside higher scores on creative thinking tests. The findings indicate that education impacts how children represent concepts in semantic memory and suggest that different educational experiences can affect higher cognitive functions, including creative thinking.

To perform counterfactual reasoning in Structural Causal Models (SCMs), one needs to know the causal mechanisms, which provide factorizations of conditional distributions into noise sources and deterministic functions mapping realizations of noise to samples. Unfortunately, the causal mechanism is not uniquely identified by data that can be gathered by observing and interacting with the world, so there remains the question of how to choose causal mechanisms. In recent work, Oberst & Sontag (2019) propose Gumbel-max SCMs, which use Gumbel-max reparameterizations as the causal mechanism due to an appealing counterfactual stability property. However, the justification requires appealing to intuition. In this work, we instead argue for choosing a causal mechanism that is best under a quantitative criteria such as minimizing variance when estimating counterfactual treatment effects. We propose a parameterized family of causal mechanisms that generalize Gumbel-max. We show that they can be trained to minimize counterfactual effect variance and other losses on a distribution of queries of interest, yielding lower variance estimates of counterfactual treatment effect than fixed alternatives, also generalizing to queries not seen at training time.

Low-rank and nonsmooth matrix optimization problems capture many fundamental tasks in statistics and machine learning. While significant progress has been made in recent years in developing efficient methods for\textit {smooth} low-rank optimization problems that avoid maintaining high-rank matrices and computing expensive high-rank SVDs, advances for nonsmooth problems have been slow paced. In this paper we consider standard convex relaxations for such problems. Mainly, we prove that under a natural\textit {generalized strict complementarity} condition and under the relatively mild assumption that the nonsmooth objective can be written as a maximum of smooth functions, the\textit {extragradient method}, when initialized with a" warm-start''point, converges to an optimal solution with rate while requiring only two\textit {low-rank} SVDs per iteration. We give a precise trade-off between the rank of the SVDs required and the radius of the ball in which we need to initialize the method. We support our theoretical results with empirical experiments on several nonsmooth low-rank matrix recovery tasks, demonstrating that using simple initializations, the extragradient method produces exactly the same iterates when full-rank SVDs are replaced with SVDs of rank that matches the rank of the (low-rank) ground-truth matrix to be recovered.

While problem-solving is central in our daily life, its underlying mechanisms remain largely unknown. Restructuration (i.e., reinterpretation and reorganization of problem-related representations) is theoretically considered as one such mechanism, yet empirical evidence supporting it is scarce. We investigated restructuration as a mechanism underlying problem-solving, using network science methodology. We estimated the structure of participant’s individual semantic memory network before and after they attempted to solve a riddle. These networks represent the organization of solution-relevant and irrelevant terms as nodes, with edges representing the strength of relationship between them based on participants’ relatedness judgments. The difference in semantic network metrics between pre- and post-solving phases was used to quantify restructuration and predict successful problem-solving. Problem-solving was predicted by local restructuration of semantic network, only in edges and nodes that had been assessed as helpful to solve the problem. These results shed new light on the mental restructuring associated with problem-solving and provide a new method to quantify this restructuring.

BACKGROUND Infection with SARS-CoV-2 provides substantial natural immunity against reinfection. Recent studies have shown strong waning of the immunity provided by the BNT162b2 vaccine. The time course of natural and hybrid immunity is unknown. METHODS Data on confirmed SARS-CoV-2 infections were extracted from the Israeli Ministry of Health database for the period August to September 2021 regarding all persons previously infected or vaccinated. We compared infection rates as a function of time since the last immunity-conferring event using Poisson regression, adjusting for possible confounding factors. RESULTS Confirmed infection rates increased according to time elapsed since the last immunity-conferring event in all cohorts. For unvaccinated previously infected individuals they increased from 10.5 per 100,000 risk-days for those previously infected 4-6 months ago to 30.2 for those previously infected over a year ago. For individuals receiving a single dose following prior infection they increased from 3.7 per 100,000 person days among those vaccinated in the past two months to 11.6 for those vaccinated over 6 months ago. For vaccinated previously uninfected individuals the rate per 100,000 person days increased from 21.1 for persons vaccinated within the first two months to 88.9 for those vaccinated more than 6 months ago. CONCLUSIONS Protection from reinfection decreases with time since previous infection, but is, nevertheless, higher than that conferred by vaccination with two doses at a similar time since the last immunity-conferring event. A single vaccine dose after infection helps to restore protection.

With advances in reinforcement learning (RL), agents are now being developed in high-stakes application domains such as healthcare and transportation. Explaining the behavior of these agents is challenging, as the environments in which they act have large state spaces, and their decision-making can be affected by delayed rewards, making it difficult to analyze their behavior. To address this problem, several approaches have been developed. Some approaches attempt to convey the global behavior of the agent, describing the actions it takes in different states. Other approaches devised local explanations which provide information regarding the agent's decision-making in a particular state. In this paper, we combine global and local explanation methods, and evaluate their joint and separate contributions, providing (to the best of our knowledge) the first user study of combined local and global explanations for RL agents. Specifically, we augment strategy summaries that extract important trajectories of states from simulations of the agent with saliency maps which show what information the agent attends to. Our results show that the choice of what states to include in the summary (global information) strongly affects people's understanding of agents: participants shown summaries that included important states significantly outperformed participants who were presented with agent behavior in a set of world-states that are likely to appear during gameplay. We find mixed results with respect to augmenting demonstrations with saliency maps (local information), as the addition of saliency maps, in the form of raw heat maps, did not significantly improve performance in most cases. However, we do find some evidence that saliency maps can help users better understand what information the agent relies on during its decision-making, suggesting avenues for future work that can further improve explanations of RL agents.

We present an optimized rerandomization design procedure for a non-sequential treatment-control experiment. Randomized experiments are the gold standard for finding causal effects in nature. But sometimes random assignments result in unequal partitions of the treatment and control group visibly seen as imbalance in observed covariates. There can additionally be imbalance on unobserved covariates. Imbalance in either observed or unobserved covariates increases treatment effect estimator error inflating the width of confidence regions and reducing experimental power. “Rerandomization” is a strategy that omits poor imbalance assignments by limiting imbalance in the observed covariates to a prespecified threshold. However, limiting this threshold too much can increase the risk of contracting error from unobserved covariates. We introduce a criterion that combines observed imbalance while factoring in the risk of inadvertently imbalancing unobserved covariates. We then use this criterion to locate the optimal rerandomization threshold based on the practitioner’s level of desired insurance against high estimator error. We demonstrate the gains of our designs in simulation and in a dataset from a large randomized experiment in education. We provide an open source R package available on CRAN named OptimalRerandExpDesigns which generates designs according to our algorithm.

We present a new generalization of the extensible bin packing with unequal bin sizes problem. In our generalization the cost of exceeding the bin size (also known as the bin capacity) depends on the index of the bin and not only on the amount in which the size of the bin is exceeded. This generalization does not satisfy the assumptions on the cost function that were used to present the existing polynomial time approximation scheme (PTAS) for the extensible bin packing with unequal bin sizes problem. In this work, we show the existence of an efficient PTAS (EPTAS) for this new generalization and thus in particular we improve the earlier PTAS for the extensible bin packing with unequal bin sizes problem into an EPTAS. Our new scheme is based on using the shifting technique followed by a solution of a polynomial number of n-fold programming instances. In addition, we present an asymptotic fully polynomial time approximation scheme for the related bin packing type variant of the problem.

A wealth of research documents the critical role of trust for social exchange and cooperative behavior. The ability to inspire trust in others can often be elusive, and distrust can have adverse interpersonal and ethical consequences. Drawing from the literature on social hierarchy and interpersonal judgments, the current research explores the predictive role of a structural paradox between high power and low status in identifying the actors most likely to be distrusted and monitored for ethical misconduct. Across four studies and an internal meta-analysis, we found that the structural paradox was associated with distrust-related judgments and behaviors. In Study 1, high power-low status actors were judged as less trustworthy. In Studies 2 and 3, high power-low status actors were sent less money in a trust game, an effect fully mediated by feelings of dislike. Study 4 revealed that high power-low status actors were more likely to be monitored for cheating, an effect partially mediated by trust judgments. These findings contribute to business ethics research by identifying the structural paradox of high power-low status as a salient contextual influence impacting observers’ distrust and monitoring dynamics. Implications for reducing observers’ level of distrust of high power-low status actors are discussed.

We consider the limiting extremal process $X$ of the particles of the binary branching Brownian motion. We show that after a shift by the logarithm of the derivative martingale Z, the rescaled “density” of particles, which are at distance $n+x$ from a position close to the tip of $X$, converges in probability to a multiple of the exponential $e^x$ as $n\to +\infty$. We also show that the fluctuations of the density, after another scaling and an additional random but explicit shift, converge to a 1-stable random variable. Our approach uses analytic techniques and is motivated by the connection between the properties of the branching Brownian motion and the Bramson shift of the solutions to the Fisher-KPP equation with some specific initial conditions initiated in [9], [10] and further developed in the present paper. The proofs of the limit theorems for $X$ rely crucially on the fine asymptotics of the behavior of the Bramson shift for the Fisher-KPP equation starting with initial conditions of “size” $0<\epsilon \ll 1$, up to terms of the order ${\left[\right(\log {\epsilon }^{-1}\left)\right]}^{-1-\gamma }$, with some $\gamma >0$.

The celebrated Bayesian persuasion model considers strategic communication between an informed agent (the sender) and uniformed decision makers (the receivers). The current rapidly-growing literature assumes a dichotomy: either the sender is powerful enough to communicate separately with each receiver (a.k.a. private persuasion), or she cannot communicate separately at all (a.k.a. public persuasion). We propose a model that smoothly interpolates between the two, by introducing a natural multi-channel communication structure in which each receiver observes a subset of the sender's communication channels. This captures, e.g., receivers on a network, where information spillover is almost inevitable. We completely characterize when one communication structure is better for the sender than another, in the sense of yielding higher optimal expected utility universally over all prior distributions and utility functions. The characterization is based on a simple pairwise relation among receivers - one receiver information-dominates another if he observes at least the same channels. We prove that a communication structure M_1 is (weakly) better than M_2 if and only if every information-dominating pair of receivers in M_1 is also such in M_2. We also provide an additive FPTAS for the optimal sender's signaling scheme when the number of states is constant and the graph of information-dominating pairs is a directed forest. Finally, we prove that finding an optimal signaling scheme under multi-channel persuasion is computationally hard for a general family of sender's utility functions that admit computationally tractable optimal signaling schemes under both public and private persuasion.

We consider a continuous-time random walk on a regular tree of finite depth and study its favorite points among the leaf vertices. For the walk started from a leaf vertex and stopped upon hitting the root we prove that, in the limit as as the depth of the tree tends to infinity, the suitably scaled and centered maximal time spent at any leaf converges to a randomly-shifted Gumbel law. The random shift is characterized using a derivative-martingale like object associated with square-root local-time process on the tree.

We present an algorithm that enumerates all the minimal triangulations of a graph in incremental polynomial time. Consequently, we get an algorithm for enumerating all the proper tree decompositions, in incremental polynomial time, where “proper” means that the tree decomposition cannot be improved by removing or splitting a bag. The algorithm can incorporate any method for (ordinary, single result) triangulation or tree decomposition, and can serve as an anytime algorithm to improve such a method. We describe an extensive experimental study of an implementation on real data from different fields. Our experiments show that the algorithm improves upon central quality measures over the underlying tree decompositions, and is able to produce a large number of high-quality decompositions.

Let $G$ be an $n$-vertex graph, and $s,t$ vertices of $G$. We present an efficient algorithm which enumerates the set of minimal $st$-separators of $G$ in ascending order of cardinality, with a delay of $O(n^{3.5})$ per separator. In particular, we present an algorithm that lists, in ascending order of cardinality, all minimal separators with at most $k$ vertices. In that case, we show that the delay of the enumeration algorithm is $O(kn^{2.5})$ per separator. Our process is based on a new method that can decide, in polynomial time, whether the set of minimal separators under certain inclusion, exclusion, and cardinality constraints is empty.

We consider the non-preemptive scheduling problem on identical machines where there is a parameter B and each machine in every unit length time interval can process up to B different jobs. The goal function we consider is the makespan minimization and we develop an EPTAS for this problem. Prior to our work a PTAS was known only for the case of one machine and constant values of B, and even the case of non-constant values of B on one machine was not known to admit a PTAS.

Insufficient exploration of one’s surroundings is at the root of many real-life problems, as demonstrated by many famous biases (e.g., the status quo bias, learned helplessness). The current work focuses on the emergence of this phenomenon at the strategy level: the tendency to under-explore the set of available choice strategies. We demonstrate that insufficient exploration of strategies can also manifest as excessive exploration between options. In such cases, interventions aimed at improving choices by reducing the costs of exploration of options are likely to fail. In Study 1, participants faced an exploration task that implies an infinite number of choice strategies and a small sub-set of (near) optimal solutions. We manipulated the amount of practice participants underwent during the first, shorter game and compared their performance in a second, longer game with an identical payoff structure. Our results show that regardless of the amount of practice, participants in all experimental groups tended to under-explore the strategy space and relied on a specific strategy that implied over-exploration of the option space. That is, under-exploration of strategies was manifested as over-exploration of options. In Study 2, we added a constraint that, on a subset of practice trials, forced participants to exploit familiar options. This manipulation almost doubled the per-trial average outcome on the comparable longer second game. This suggests that forcing participants to experience the effects of different (underexplored) strategy components during practice can greatly increase the chance they make better choices later on.

Purpose: The goal of this work is to use multi-camera video to classify open surgery tools as well as identify which tool is held in each hand. Multi-camera systems help prevent occlusions in open surgery video data. Furthermore, combining multiple views such as a Top-view camera covering the full operative field and a Close-up camera focusing on hand motion and anatomy, may provide a more comprehensive view of the surgical workflow. However, multi-camera data fusion poses a new challenge: a tool may be visible in one camera and not the other. Thus, we defined the global ground truth as the tools being used regardless their visibility. Therefore, tools that are out of the image should be remembered for extensive periods of time while the system responds quickly to changes visible in the video. Methods: Participants (n=48) performed a simulated open bowel repair. A Top-view and a Close-up cameras were used. YOLOv5 was used for tool and hand detection. A high frequency LSTM with a 1 second window at 30 frames per second (fps) and a low frequency LSTM with a 40 second window at 3 fps were used for spatial, temporal, and multi-camera integration. Results: The accuracy and F1 of the six systems were: Top-view (0.88/0.88), Close-up (0.81,0.83), both cameras (0.9/0.9), high fps LSTM (0.92/0.93), low fps LSTM (0.9/0.91), and our final architecture the Multi-camera classifier(0.93/0.94). Conclusion: By combining a system with a high fps and a low fps from the multiple camera array we improved the classification abilities of the global ground truth.

In IVF treatments, extended culture to single blastocyst-transfer is the recommended protocol over cleavage-stage transfer. However, evidence-based criteria for assessing the heterogeneous implications on implantation outcome are lacking. To estimate the causal effect of blastocyst-transfer on implantation outcome, we assembled a multicenter dataset of embryo time-lapse imaging. The data includes a natural source of randomness and has a strong claim for satisfying the assumptions needed for valid causal inference. By fitting a causal forest model, we assessed the ‘Transfer Lift’, which quantifies the probability difference in embryo implantation if transferred as a blastocyst versus cleavage-stage. Blastocyst transfer increased the average implantation rate, however we revealed a subpopulation of negative Transfer Lift embryos whose implantation potential is predicted to increase via cleavage-stage transfer. We provide day-of-transfer decision-support tools that are retrospectively estimated to improve implantation rate by 32%, thus demonstrating the efficacy of embryo-level causal inference in reproductive medicine.

We consider the Ordered Open End Bin Packing problem. Items of sizes in (0, 1] are presented one by one, to be assigned to bins in this order. An item can be assigned to any bin for which the current total size is strictly below 1. This means also that the bin can be overloaded by its last packed item. We improve lower and upper bounds on the asymptotic competitive ratio in the online case. Specifically, we design the first algorithm whose asymptotic competitive ratio is strictly below 2, and its value is close to the lower bound. This is in contrast to the best possible absolute competitive ratio, which is equal to 2. We also study the offline problem where the sequence of items is known in advance, while items are still assigned to bins based on their order in the sequence. For this scenario, we design an asymptotic polynomial time approximation scheme.

We consider extensive form 2-player win-lose games, with alternating moves, of perfect and complete information. The games are played over a complete binary-tree of depth n, where 0/1 payoffs in the leaves are drawn according to an i.i.d. Bernoulli distribution with probability p. Whenever p differs from the golden ratio, asymptotically as $n\to \infty$, the winner of the game is determined. In the case where p equals the golden ratio, we call such a random game a golden game. In golden games the winner is the player that acts first with probability equal to the golden ratio. We suggest the notion of fragility as a measure for “fairness” of a game's rules. Fragility counts how many leaves' payoffs should be flipped in order to convert the identity of the winning player. Our main result provides a recursive formula for asymptotic fragility of golden games. Surprisingly, golden games are extremely fragile. For instance, with probability ≈0.77 a losing player could flip a single payoff (out of $2^n$) and become a winner. With probability ≈0.999 a losing player could flip 3 payoffs and become the winner.

Estimating personalized treatment effects from high-dimensional observational data is essential in situations where experimental designs are infeasible, unethical, or expensive. Existing approaches rely on fitting deep models on outcomes observed for treated and control populations. However, when measuring individual outcomes is costly, as is the case of a tumor biopsy, a sample-efficient strategy for acquiring each result is required. Deep Bayesian active learning provides a framework for efficient data acquisition by selecting points with high uncertainty. However, existing methods bias training data acquisition towards regions of non-overlapping support between the treated and control populations. These are not sample-efficient because the treatment effect is not identifiable in such regions. We introduce causal, Bayesian acquisition functions grounded in information theory that bias data acquisition towards regions with overlapping support to maximize sample efficiency for learning personalized treatment effects. We demonstrate the performance of the proposed acquisition strategies on synthetic and semi-synthetic datasets IHDP and CMNIST and their extensions, which aim to simulate common dataset biases and pathologies.

Rationale

Methylphenidate (MPH) and mixed D,L-amphetamine salts (MASs; Adderall) were previously found to have unreliable effects on judgment and decision processes.

Objective

We predicted that MPH and MASs have a specific effect of reducing heuristic responses, which should lead to increased performance on the cognitive reflection test (CRT). The CRT is considered to be a testbed for heuristic versus deliberative response modes.

Methods

We recruited a sample of 15,361 individuals using the Prolific Academic crowdsourcing platform. From this initial pool, our final sample consisted of 294 participants (125 MPH users and 169 MASs users) who conformed to the study criteria and completed the experimental tasks. Tasks were performed on days where participants were either medicated or not, allowing to assess the effect of medication status.

Results

There was a strong positive effect of taking MPH on CRT scores (Cohen’s d = 0.40) which was not qualified by frequency of MPH usage, ADHD symptoms, and demographic factors. There was also a somewhat weaker effect for MASs (Cohen’s d = 0.07). No effects of MPH and MASs were recorded for risk-taking and numeracy.

Conclusions

The results indicate that MPH enhances decision-making in tasks where heuristic responses typically bias it.

In this work, we prove that for any dimension $\mathit{d}\ge 1$ and any $\mathit{\gamma }\in (0,1)$ super-Brownian motion corresponding to the log-Laplace equation

$$\mathit{v}(\mathit{t},\mathit{x})=({\mathit{S}}_{\mathit{t}}\mathit{f})(\mathit{x})-{\int }_0^{\mathit{t}}({\mathit{S}}_{\mathit{t}-\mathit{s}}{\mathit{v}}^{\mathit{\gamma }}(\mathit{s},·))(\mathit{x})\mathrm{d}\mathit{s},(\mathit{t},\mathit{x})\in {\mathbb{R}}_{+}\times {\mathbb{R}}^{\mathit{d}},$$

is absolutely continuous with respect to Lebesgue measure at any fixed time $\mathit{t}>0$. ${\{{\mathit{S}}_{\mathit{t}}\}}_{\mathit{t}\ge 0}$ denotes a transition semigroup of a standard Brownian motion. Our proof is based on properties of solutions of the log-Laplace equation. We also prove that when initial datum $\mathit{v}(0,·)$ is a finite, non-zero measure, then the log-Laplace equation has a unique, continuous solution. Moreover this solution continuously depends on initial data.

We advance the theoretical and practical understanding of affect in service interactions by conceptualizing employees and customers as concurrent participants in the same interaction. We analyzed employees' emotional labor requirements, which comprise both the well-recognized requirement to display positive affect (i.e., acting is response independent) and the less-recognized requirement to attend to customers' affective displays (i.e., acting is response dependent). We found support for our hypotheses across two studies, which compare the affective displays of employees and customers within the same interactions. In Study 1, we examined field data comprising 1 320 392 customer and employee messages from 164 899 real-life chat-based service interactions and used automated sentiment analysis to identify displays of positive and negative affect. In Study 2, we used simulated service interactions to examine discrete emotions. Using different methodologies, both Studies 1 and 2 found that employees and customers differ in their response-dependent affective behaviors. Study 2 also demonstrated that employee response-dependent affective behavior improves customer outcomes. Our analyses enrich the understanding of emotional labor, contribute to the theory of affect in social interactions by considering concurrent yet differing affective behaviors of partners in the same interaction, and suggest new and exciting methods for Organizational Behavior research.

With the onset of the coronavirus pandemic, clinicians, public officials, and citizens alike struggled to stay abreast of the constant and evolving stream of information about the clinical manifestations of illness, epidemiology of the disease, and the public health response. In this paper, we adapted (ie, added and modified elements) Beasley and colleagues’ information chaos framework to understand the context of citizens’ experiences with information during the COVID-19 pandemic. We will show how our adapted framework can be used to characterize information associated challenges observed during this time and the possible impact of information chaos on peoples’ cognition and behaviours. Ultimately, we believe that research will benefit by adopting a more holistic perspective using the information chaos framework than strictly studying the independent factors in isolation.

Location classification is used for associating type to locations, to enrich maps and support a plethora of geospatial applications that rely on location types. Classification can be performed by humans, but using machine learning is more efficient and faster to react to changes than human-based classification. Machine learning can be used in lieu of human classification or for supporting it. In this paper we study the use of machine learning for Geosocial Location Classification, where the type of a site, e.g., a building, is discovered based on social-media posts, e.g., tweets. Our goal is to correctly associate a set of tweets posted in a small radius around a given location with the corresponding location type, e.g., school, church, restaurant or museum. We explore two approaches to the problem: (a) a pipeline approach, where each post is first classified, and then the location associated with the set of posts is inferred from the individual post labels; and (b) a joint approach where the individual posts are simultaneously processed to yield the desired location type. We tested the two approaches over a data set of geotagged tweets. Our results demonstrate the superiority of the joint approach. Moreover, we show that due to the unique structure of the problem, where weakly-related messages are jointly processed to yield a single final label, linear classifiers outperform deep neural network alternatives.

Our goal in this paper is to connect workspace design to employee wellbeing and social sustainability. Toward this connection, we introduce and empirically test a new concept of “workspace integration”. This concept refers to the continuum of integration of an employee’s workspace with the organizational, physical space. We further define three workspace affordances that predict the social sustainability of workspace arrangements by influencing employee wellbeing. The three affordances are perceptions of symbolism of the organization and of symbolism of the self, and opportunities for social interactions. We evaluate our theory using data collected from an online survey of British employees (n = 392) working in offices at home and/or in the organization in various industries. As predicted, workspace integration positively influences wellbeing directly as well as indirectly via mediation of symbolism of the organization and opportunities for social interactions. The third affordance—symbolism of the self—positively influences wellbeing, but is not affected by workspace integration. Our findings confirm the impact of workspace affordances on employee wellbeing, and thus their utility for the analysis of social sustainability. The findings also contribute to understanding of the differences between workspace symbolism of the organization and of the self, their impact on wellbeing, and thus their implications for social sustainability.

External enforcement policies aimed to reduce violations differ on two key components: the probability of inspection and the severity of the punishment. Different lines of research offer different insights regarding the relative importance of each component. In four studies, students and Prolific crowdsourcing participants (Ntotal = 816) repeatedly faced temptations to commit violations under two enforcement policies. Controlling for expected value, we found that a policy combining a high probability of inspection with a low severity of fines (HILS) was more effective than an economically equivalent policy that combined a low probability of inspection with a high severity of fines (LIHS). The advantage of prioritizing inspection frequency over punishment severity (HILS over LIHS) was greater for participants who, in the absence of enforcement, started out with a higher violation rate. Consistent with studies of decisions from experience, frequent enforcement with small fines was more effective than rare severe fines even when we announced the severity of the fine in advance to boost deterrence. In addition, in line with the phenomenon of underweighting of rare events, the effect was stronger when the probability of inspection was rarer (as in most real-life inspection probabilities) and was eliminated under moderate inspection probabilities. We thus recommend that policymakers looking to effectively reduce recurring violations among noncriminal populations should consider increasing inspection rates rather than punishment severity.

Industry 4.0 offers opportunities to combine multiple sensor data sources using IoT technologies for better utilization of raw material in production lines. A common belief that data is readily available (the big data phenomenon), is oftentimes challenged by the need to effectively acquire quality data under severe constraints. In this paper we propose a design methodology, using active learning to enhance learning capabilities, for building a model of production outcome using a constrained amount of raw material training data. The proposed methodology extends existing active learning methods to effectively solve regression-based learning problems and may serve settings where data acquisition requires excessive resources in the physical world. We further suggest a set of qualitative measures to analyze learners performance. The proposed methodology is demonstrated using an actual application in the milk industry, where milk is gathered from multiple small milk farms and brought to a dairy production plant to be processed into cottage cheese.

In competitive search settings such as the Web, many documents' authors (publishers) opt to have their documents highly ranked for some queries. To this end, they modify the documents --- specifically, their content --- in response to induced rankings. Thus, the search engine affects the content in the corpus via its ranking decisions. We present a first study of the ability of search engines to drive pre-defined, targeted, content effects in the corpus using simple techniques. The first is based on the herding phenomenon --- a celebrated result from the economics literature --- and the second is based on biasing the relevance ranking function. The types of content effects we study are either topical or touch on specific document properties --- length and inclusion of query terms. Analysis of ranking competitions we organized between incentivized publishers shows that the types of content effects we target can indeed be attained by applying our suggested techniques. These findings have important implications with regard to the role of search engines in shaping the corpus.

The goal of this study was to develop new reliable open surgery suturing simulation system for training medical students in situation where resources are limited or in the domestic setup. Namely, we developed an algorithm for tools and hands localization as well as identifying the interactions between them based on simple webcam video data, calculating motion metrics for assessment of surgical skill. Twenty-five participants performed multiple suturing tasks using our simulator. The YOLO network has been modified to a multi-task network, for the purpose of tool localization and tool-hand interaction detection. This was accomplished by splitting the YOLO detection heads so that they supported both tasks with minimal addition to computer run-time. Furthermore, based on the outcome of the system, motion metrics were calculated. These metrics included traditional metrics such as time and path length as well as new metrics assessing the technique participants use for holding the tools. The dual-task network performance was similar to that of two networks, while computational load was only slightly bigger than one network. In addition, the motion metrics showed significant differences between experts and novices. While video capture is an essential part of minimally invasive surgery, it is not an integral component of open surgery. Thus, new algorithms, focusing on the unique challenges open surgery videos present, are required. In this study, a dual-task network was developed to solve both a localization task and a hand-tool interaction task. The dual network may be easily expanded to a multi-task network, which may be useful for images with multiple layers and for evaluating the interaction between these different layers.

We consider the problem of using expert data with unobserved confounders for imitation and reinforcement learning. We begin by defining the problem of learning from confounded expert data in a contextual MDP setup. We analyze the limitations of learning from such data with and without external reward, and propose an adjustment of standard imitation learning algorithms to fit this setup. We then discuss the problem of distribution shift between the expert data and the online environment when the data is only partially observable. We prove possibility and impossibility results for imitation learning under arbitrary distribution shift of the missing covariates. When additional external reward is provided, we propose a sampling procedure that addresses the unknown shift and prove convergence to an optimal solution. Finally, we validate our claims empirically on challenging assistive healthcare and recommender system simulation tasks.

Many real-life choices are based on previous experiences. Research devoted to these decisions from experience has typically employed static settings, where the probability of a given outcome is constant across trials. However, recent studies of repeated choice suggest that people tend to follow perceived patterns of outcomes even when true patterns do not exist (i.e., in static settings). Here, we examine whether the tendency to follow perceived patterns above and beyond external incentives also characterizes decisions from sampling. To this aim, we modified the static sampling paradigm to include a fixed sequence of outcomes while the incentive dictate disregarding the sequence. In two studies, we found a strong tendency to follow the fixed pattern of outcomes. This tendency was evident not only in sampling choices where following the pattern required additional effort and did not provide additional information. The same tendency was also evident in participant’s final consequential choices, where following the pattern impaired financial returns. The results were replicated after ensuring comprehension of the task, doubling the expected payoffs, and under a partial-feedback design. Overall, our results suggest that decisions from sampling, like repeated consequential choice, reflect a strong tendency to follow perceived environmental regularities. Our results are consistent with the assumption that during free sampling and during consequential choice, most participants respond to when (i.e., on which specific trials) each of the options is better rather than to which option is better overall (i.e., implies a more attractive distribution of outcomes).

Preemptive scheduling problems of minimizing the makespan on parallel machines are basic problems. Motivated by sensitivity analysis and online algorithms, we investigate the problem of designing robust algorithms that are faced with the input one job at a time, but unlike online algorithms they are allowed to modify a small portion of the solution whenever a new job is released. For the settings of uniformly related machines, it is known that such algorithms cannot maintain an optimal solution (for general input sequences). Thus, we consider the semi-online case where the job sequence has non-increasing job sizes and design an optimal algorithm with migration factor 1. Previously such result was known only for the case of identical machines where it holds for an arbitrary input sequence.

Objective

We define and demonstrate the use of OPM-TA—a model-based task analysis (TA) framework that uses object-process methodology (OPM) ISO 19450 as a viable alternative to traditional TA techniques.

Background

A variety of different TA methods exist in human factors engineering, and several of them are often applied successively for a broad task representation, making it difficult to follow.

Method

Using OPM-TA, we modeled how an International Space Station (ISS) astronaut would support extravehicular activities using the existing robotic arm workstation with a new control panel and an electronic procedure system. The modeling employed traditional TA methods and the new OPM-TA approach, enabling a comparison between them.

Results

While the initial stages of modeling with OPM-TA follow those of traditional TA, OPM-TA modeling yields an executable and logically verifiable model of the entire human–robot system. Both OPM’s hierarchical set of diagrams and the equivalent, automatically generated statements in a subset of natural language text specify how objects and processes relate to each other at increasingly detailed levels. The graphic and textual OPM modalities specify the system’s architecture, which enables its function and benefits its users. To verify the model logical correctness model, we executed it using OPM’s simulation capability.

Conclusion

OPM-TA was able to unify traditional TA methods and expand their capabilities. The formal yet intuitive OPM-TA approach fuses and extends traditional TA methods, which are not amenable to simulation. It therefore can potentially become a widely used means for TA and human–machine procedure development and testing.

Objective To analyze the correlation between COVID-19 vaccination percentage and socioeconomic status (SES). Methods A nationwide ecologic study based on open-sourced, anonymized, aggregated data provided by the Israel Ministry of Health. The correlations between municipal SES, vaccination percentage, and active COVID-19 cases during the vaccination campaign were analyzed by using weighted Pearson correlations. To assess the adequacy of first dose vaccination rollout relative to the municipality COVID-19 disease burden, a metric termed the vaccination need ratio was devised by dividing the total number of active cases (per 10,000 people) by the vaccination percentage of the population over 60 in each municipality, and its correlation with the SES was examined. Results 23 days after initiation of the vaccination campaign, 760,916 (56.8%) individuals over the age of 60 were vaccinated in Israel with the first dose of the BNT162b2 COVID-19 vaccine. A negative correlation was found between the COVID-19 active case burden and the vaccination percentage of the study population in each municipality (r=-0.47, 95% confidence interval [-0.59 to -0.30]). The vaccination percentage significantly correlated with the municipal SES (r=0.83, 95% confidence interval [0.79 to 0.87]). This finding persisted but was attenuated over a five-week period. A negative correlation between the vaccination need ratio and municipal SES (r=-0.80, 95% confidence interval [-0.88 to -0.66]) was found. Conclusion Lower COVID-19 vaccination percentage was associated with lower SES and high active disease burden. Vaccination efforts should focus on areas with lower SES and high disease burden to assure equality of vaccine allocation and potentially provide a more diligent disease mitigation.

Advances in automation have provided for sustained productivity increases and manufacturing growth over the past decade. Sustaining this growth will require automation to become more agile and flexible, enabling the automation of tasks that require a high degree of human dexterity and the ability to react to unforeseen circumstances. Applying robots is one promising approach, but their traditional program-by-teaching model takes considerable time, requires extensive expertise, and does not lend itself to tasks that require adaptability. This has limited robots to high-volume, repetitive operations and precluded them from low-volume, time critical, and flexible projects. Off-line programming of robots is possible, similar to the computer-aided manufacturing (CAM) method widely used for machine tools. However, the poor accuracy of robots compared with machine tools limits them to jobs with low tolerance requirements, or requires additional methods such as calibration, modeling, and external sensing to improve their accuracy. These methods increase the upfront cost of a robotic system. However, advances and cost reduction in sensing technologies (especially laser scanning) have brought robot systems into the price range of even small-to-medium enterprises. In addition, use of end-of-arm tools (EOAT) has given integrators the ability to provide faster turnaround time and utilize the same infrastructure in a high-mix, low-volume environment.

The problem of an autonomous agent moving on a planar surface, such as an aerial drone or a small naval vessel can be treated as navigation between a series of points. While nominally the movement between each pair of points can be treated as a 1D projection of the movement on the vector connecting the two points, in the presence of arbitrary constant disturbance the full problem on a plane must be considered. The minimum-time optimal solution is now dependent on the value and direction of the disturbance, which naturally affects the completion of the movement task. In this work, we address the problem of minimum time movement on a 2D plane with quadratic drag, under norm state (velocity), and norm control (acceleration) constraints. The structure and properties of the optimal solution of this nonlinear problem are found and analyzed. The Pontryagin Maximum Principle (PMP) with control and state constraints is utilized. Simulations encouraging the results are presented and compared with those of the academic open-source optimal control solver Falcon.m.

We consider fair allocation of indivisible items in a model with goods, chores, and copies, as a unified framework for studying: (1)~the existence of EFX and other solution concepts for goods with copies; (2)~the existence of EFX and other solution concepts for chores. We establish a tight relation between these issues via two conceptual contributions: First, a refinement of envy-based fairness notions that we term envy \emph{without commons} (denoted $\efxwc$ when applied to EFX). Second, a formal \emph{duality theorem} relating the existence of a host of (refined) fair allocation concepts for copies to their existence for chores. We demonstrate the usefulness of our duality result by using it to characterize the existence of EFX for chores through the dual environment, as well as to prove EFX existence in the special case of leveled preferences over the chores. We further study the hierarchy among envy-freeness notions without commons and their $\alpha$-MMS guarantees, showing for example that any $\efxwc$ allocation guarantees at least $\frac{4}{11}$-MMS for goods with copies.

Weighted voting games apply to a wide variety of multi-agent settings. They enable the formalization of power indices which quantify the coalitional power of players. We take a novel approach to the study of the power of big vs. small players in these games. We model small (big) players as having single (multiple) votes. The aggregate relative power of big players is measured w.r.t. their votes proportion. For this ratio, we show small constant worst-case bounds for the Shapley-Shubik and the Deegan-Packel indices. In sharp contrast, this ratio is unbounded for the Banzhaf index. As an application, we define a false-name strategic normal form game where each big player may split its votes between false identities, and study its various properties. Together, our results provide foundations for the implications of players’ size, modeled as their ability to split, on their relative power.

In reactive synthesis, one begins with a temporal specification \(\varphi \), and automatically synthesizes a system \(M\) such that \(M\models \varphi \). As many systems can satisfy a given specification, it is natural to seek ways to force the synthesis tool to synthesize systems that are of a higher quality, in some well-defined sense. In this article we focus on a well-known measure of the way in which a system satisfies its specification, namely vacuity. Our conjecture is that if the synthesized system M satisfies \(\varphi \) non-vacuously, then M is likely to be closer to the user’s intent, because it satisfies \(\varphi \) in a more “meaningful” way. Narrowing the gap between the formal specification and the designer’s intent in this way, automatically, is the topic of this article. Specifically, we propose a bounded synthesis method for achieving this goal. The notion of vacuity as defined in the context of model checking, however, is not necessarily refined enough for the purpose of synthesis. Hence, even when the synthesized system is technically non-vacuous, there are yet more interesting (equivalently, less vacuous) systems, and we would like to be able to synthesize them. To that end, we cope with the problem of synthesizing a system that is as non-vacuous as possible, given that the set of interesting behaviours with respect to a given specification induce a partial order on transition systems. On the theoretical side we show examples of specifications for which there is a single maximal element in the partial order (i.e., the most interesting system), a set of equivalent maximal elements, or a number of incomparable maximal elements. We also show examples of specifications that induce infinite chains of increasingly interesting systems. These results have implications on how non-vacuous the synthesized system can be. We implemented the new procedure in our synthesis tool PARTY. For this purpose we added to it the capability to synthesize a system based on a property which is a conjunction of universal and existential LTL formulas.

Cognitive researchers often carve cognition up into structures and processes. Cognitive processes operate on structures, like vehicles driving over a map. Language alongside semantic and episodic memory are proposed to have structure, as are perceptual systems. Over these structures, processes operate to construct memory and solve problems by retrieving and manipulating information. Network science offers an approach to representing cognitive structures and has made tremendous inroads into understanding the nature of cognitive structure and process. But is the mind a network? If so, what kind? In this article, we briefly review the main metaphors, assumptions, and pitfalls prevalent in cognitive network science (maps and vehicles; one network/process to rule them all), highlight the need for new metaphors that elaborate on the map-and-vehicle framework (wormholes, skyhooks, and generators), and present open questions in studying the mind as a network (the challenge of capturing network change, what should the edges of cognitive networks be made of, and aggregated vs. individual-based networks). One critical lesson of this exercise is that the richness of the mind as network approach makes it a powerful tool in its own right; it has helped to make our assumptions more visible, generating new and fascinating questions, and enriching the prospects for future research. A second lesson is that the mind as a network–though useful–is incomplete. The mind is not a network, but it may contain them.

Several recent studies have explored the interaction effects between topics, systems, corpora, and components when measuring retrieval effectiveness. However, all of these previous studies assume that a topic or information need is represented by a single query. In reality, users routinely reformulate queries to satisfy an information need. Recently there has been renewed interest in the notion of “query variations” which are essentially multiple user formulations for an information need. Like many retrieval models, some queries are highly effective while others are not. In this work, we explore the fundamental problem of studying the interaction components of an IR experimental collection. Our findings show that query formulations have a comparable effect size to the topic factor itself, which is known to be the factor with the greatest effect size in prior ANOVA studies. This suggests that topic difficulty is an artifact of the collection considered and highlights the importance of further research in understanding link between the complexity of a topic and the query rewriting in IR related tasks

Schema matching is a core task of any data integration process. Being investigated in the fields of databases, AI, Semantic Web and data mining for many years, the main challenge remains the ability to generate quality matches among data concepts (e.g., database attributes). In this work, we examine a novel angle on the behavior of humans as matchers, studying match creation as a process. We analyze the dynamics of common evaluation measures (precision, recall, and f-measure), with respect to this angle and highlight the need for unbiased matching to support this analysis. Unbiased matching, a newly defined concept that describes the common assumption that human decisions represent reliable assessments of schemata correspondences, is, however, not an inherent property of human matchers. In what follows, we design PoWareMatch that makes use of a deep learning mechanism to calibrate and filter human matching decisions adhering the quality of a match, which are then combined with algorithmic matching to generate better match results. We provide an empirical evidence, established based on an experiment with more than 200 human matchers over common benchmarks, that PoWareMatch predicts well the benefit of extending the match with an additional correspondence and generates high quality matches. In addition, PoWareMatch outperforms state-of-the-art matching algorithms.

We consider the vector partition problem, where n agents, each with a d-dimensional attribute vector, are to be partitioned into p parts so as to minimize cost which is a given function on the sums of attribute vectors in each part. The problem has applications in a variety of areas including clustering, logistics and health care. We consider the complexity and parameterized complexity of the problem under various assumptions on the natural parameters p,d,a,t of the problem where a is the maximum absolute value of any attribute and t is the number of agent types, and raise some of the many remaining open problems.

This paper studies the minimization of a broad class of nonsmooth nonconvex objective functions subject to nonlinear functional equality constraints, where the gradients of the differentiable parts in the objective and the constraints are only locally Lipschitz continuous. We propose a specific proximal linearized alternating direction method of multipliers in which the proximal parameter is generated dynamically, and we design an explicit and tractable backtracking procedure to generate it. We prove subsequent convergence of the method to a critical point of the problem, and global convergence when the problem’s data are semialgebraic. These results are obtained with no dependency on the explicit manner in which the proximal parameter is generated. As a byproduct of our analysis, we also obtain global convergence guarantees for the proximal gradient method with a dynamic proximal parameter under local Lipschitz continuity of the gradient of the smooth part of the nonlinear sum composite minimization model.

The cross entropy loss is widely used due to its effectiveness and solid theoretical grounding. However, as training progresses, the loss tends to focus on hard to classify samples, which may prevent the network from obtaining gains in performance. While most work in the field suggest ways to classify hard negatives, we suggest to strategically leave hard negatives behind, in order to focus on misclassified samples with higher probabilities. We show that adding to the optimization goal the expectation loss, which is a better approximation of the zero-one loss, helps the network to achieve better accuracy. We, therefore, propose to shift between the two losses during training, focusing more on the expectation loss gradually during the later stages of training. Our experiments show that the new training protocol improves performance across a diverse set of classification domains, including computer vision, natural language processing, tabular data, and sequences. Our code and scripts are available at supplementary.

The classical herding model examines the asymptotic behavior of agents who observe their predecessors' actions as well as a private signal from an exogenous information structure. In this paper, we introduce a self-interested sender into the model and study her problem of designing this information structure. If agents cannot observe each other the model reduces to the Bayesian persuasion framework. However, when the agents do observe predecessors' actions, they may learn from each other, possibly lowering the sender's utility. We identify necessary and sufficient conditions under which the sender can nevertheless obtain the same utility as when the agents are unable to observe each other.

The problem considered is the non-preemptive scheduling of independent jobs that consume a resource (which is non-renewable and replenished regularly) on parallel uniformly related machines. The input defines the speed of machines, size of jobs, the quantity of resource required by the jobs, the replenished quantities, and replenishment dates of the resource. Every job can start processing only after the required quantity of the resource is allocated to the job. The objective function is the minimization of the convex combination of the makespan and an objective that is equivalent to the l_p-norm of the vector of loads of the machines. We present an EPTAS for this problem. Prior to our work only a PTAS was known in this non-renewable resource settings and this PTAS was only for the special case of our problem of makespan minimization on identical machines.

A fundamental goal of scientific research is to learn about causal relationships. However, despite its critical role in the life and social sciences, causality has not had the same importance in Natural Language Processing (NLP), which has traditionally placed more emphasis on predictive tasks. This distinction is beginning to fade, with an emerging area of interdisciplinary research at the convergence of causal inference and language processing. Still, research on causality in NLP remains scattered across domains without unified definitions, benchmark datasets and clear articulations of the remaining challenges. In this survey, we consolidate research across academic areas and situate it in the broader NLP landscape. We introduce the statistical challenge of estimating causal effects, encompassing settings where text is used as an outcome, treatment, or as a means to address confounding. In addition, we explore potential uses of causal inference to improve the performance, robustness, fairness, and interpretability of NLP models. We thus provide a unified overview of causal inference for the computational linguistics community.

The influence vanishing property in social networks states that the influence of the most influential agent vanishes as society grows. Removing this assumption causes a failure of learning of boundedly rational dynamics. We suggest a boundedly rational methodology that leads to learning in almost all networks. The methodology adjusts the agent's weights based on the Sinkhorn-Knopp matrix scaling algorithm. It is a simple, local, Markovian, and time-independent methodology that can be applied to multiple settings.

How does aging affect facial attractiveness? We tested the hypothesis that people find older faces less attractive than younger faces, and furthermore, that these aging effects are modulated by the age and sex of the perceiver and by the specific kind of attractiveness judgment being made. Using empirical and computational network science methods, we confirmed that with increasing age, faces are perceived as less attractive. This effect was less pronounced in judgments made by older than younger and middle-aged perceivers, and more pronounced by men (especially for female faces) than women. Attractive older faces were perceived as elegant more than beautiful or gorgeous. Furthermore, network analyses revealed that older faces were more similar in attractiveness and were segregated from younger faces. These results indicate that perceivers tend to process older faces categorically when making attractiveness judgments. Attractiveness is not a monolithic construct. It varies by age, sex, and the dimensions of attractiveness being judged.

The rise of pre-trained language models has yielded substantial progress in the vast majority of Natural Language Processing (NLP) tasks. However, a generic approach towards the pre-training procedure can naturally be sub-optimal in some cases. Particularly, fine-tuning a pre-trained language model on a source domain and then applying it to a different target domain, results in a sharp performance decline of the eventual classifier for many source-target domain pairs. Moreover, in some NLP tasks, the output categories substantially differ between domains, making adaptation even more challenging. This, for example, happens in the task of aspect extraction, where the aspects of interest of reviews of, e.g., restaurants or electronic devices may be very different. This paper presents a new fine-tuning scheme for BERT, which aims to address the above challenges. We name this scheme DILBERT: Domain Invariant Learning with BERT, and customize it for aspect extraction in the unsupervised domain adaptation setting. DILBERT harnesses the categorical information of both the source and the target domains to guide the pre-training process towards a more domain and category invariant representation, thus closing the gap between the domains. We show that DILBERT yields substantial improvements over state-of-the-art baselines while using a fraction of the unlabeled data, particularly in more challenging domain adaptation setups.

Recent advancements in Information Technology have provided an opportunity to significantly improve the effectiveness of inventory systems. The use of in-cycle demand information enables faster reaction to demand fluctuations. In particular, for the newsvendor (NV) system, we exploit the newly available data to perform an additional review (AR) of inventory at an endogenously determined, a priori set time during the sales period, and perform an additional replenishment if necessary. We implemented our innovative model at a market-leading media group. The results of the initial pilot were dramatic, indicating that the proposed model achieves an increase of 4%–24% in profits compared to the policy before implementation. As a result, the company started following the proposed model for all their printed magazines and observed a significant reduction in operational costs. In a generalized setting, we provide a tractable search-based optimization algorithm, based on the problem's structural properties, for determining the optimal initial quantity, AR timing, and quantity to restock at that time. Based on these theoretical results, we propose a simple heuristic that can be used for many practical situations including our implementation at Yedioth. Through a computational experiment, we show that our algorithm finds the optimal solution quickly and that the proposed heuristic performs well. We also provide additional insights into the problem—for instance, that our system exhibits properties similar to inventory pooling, provided that the demand rate is large enough.

We study the set of possible joint posterior belief distributions of a group of agents who share a common prior regarding a binary state and who observe some information structure. For two agents, we introduce a quantitative version of Aumann’s agreement theorem and show that it is equivalent to a characterization of feasible distributions from a 1995 work by Dawid and colleagues. For any number of agents, we characterize feasible distributions in terms of a “no-trade” condition. We use these characterizations to study information structures with independent posteriors. We also study persuasion problems with multiple receivers, exploring the extreme feasible distributions.

Various hard real-time systems have a desired requirement which is impossible to fulfill: to solve a computationally hard optimization problem within a short and fixed amount of time T, e.g., T = 0.5 seconds. For such a task, the exact, exponential algorithms, as well as various Polynomial-Time Approximation Schemes, are irrelevant because they can exceed T. What is left in practice is to combine various anytime algorithms in a parallel portfolio. The question is how to build such an optimal portfolio, given a budget of K computing cores. It is certainly not as simple as choosing the K best performing algorithms, because their results are possibly correlated (e.g., there is no point in choosing two good algorithm for the portfolio if they win on a similar set of instances). We prove that the decision variant of this problem is NP-complete, and furthermore that the optimization problem is approximable. On the practical side, our main contribution is a solution of the optimization problem of choosing K algorithms out of n, for a machine with K computing cores, and the related problem of detecting the minimum number of required cores to achieve an optimal portfolio, with respect to a given training set of instances. As a benchmark, we took instances of a hard optimization problem that is prevalent in the real-time industry, in which the challenge is to decide on the best action within time T. We include the results of numerous experiments that compare the various methods. Hence, a side effect of our tests is that it gives the first systematic empirical evaluation of the relative success of various known stochastic-search algorithms in coping with a hard combinatorial optimization problems under a very short and fixed timeout.

Cognitive researchers often carve cognition up into structures and processes. Cognitive processes operate on structures, like vehicles driving over a map. Language alongside semantic and episodic memory are proposed to have structure, as are perceptual systems. Over these structures, processes operate to construct memory and solve problems by retrieving and manipulating information. Network science offers an approach to representing cognitive structures and has made tremendous inroads into understanding the nature of cognitive structure and process. But is the mind a network? If so, what kind? In this article, we briefly review the main metaphors, assumptions, and pitfalls prevalent in cognitive network science (maps and vehicles; one network/process to rule them all), highlight the need for new metaphors that elaborate on the map-and-vehicle framework (wormholes, skyhooks, and generators), and present open questions in studying the mind as a network (the challenge of capturing network change, what should the edges of cognitive networks be made of, and aggregated vs. individual-based networks). One critical lesson of this exercise is that the richness of the mind as network approach makes it a powerful tool in its own right; it has helped to make our assumptions more visible, generating new and fascinating questions, and enriching the prospects for future research. A second lesson is that the mind as a network-though useful-is incomplete. The mind is not a network, but it may contain them.

We consider batched bin packing. Items are presented in a constant number of batches, and each batch should be packed before the next batch is presented. The cases of two, three, and four batches are studied. We prove improved lower bounds for the standard and parametric variants in some of the cases, and shorten the proofs for all other cases. To achieve this, we apply a new technique in our analysis, which differs from the ones previously used for proving such results.

We consider a continuous time random walk on the rooted binary tree of depth n with all transition rates equal to one and study its cover time, namely the time until all vertices of the tree have been visited. We prove that, normalized by $2^{n+1}n$ and then centered by $(\log 2)n-\log n$, the cover time admits a weak limit as the depth of the tree tends to infinity. The limiting distribution is identified as that of a Gumbel random variable with rate one, shifted randomly by the logarithm of the sum of the limits of the derivative martingales associated with two negatively correlated discrete Gaussian free fields on the infinite version of the tree. The existence of the limit and its overall form were conjectured in the literature. Our approach is quite different from those taken in earlier works on this subject and relies in great part on a comparison with the extremal landscape of the discrete Gaussian free field on the tree.

Weighted voting games apply to a wide variety of multi-agent settings. They enable the formalization of power indices which quantify the coalitional power of players. We take a novel approach to the study of the power of big vs.~small players in these games. We model small (big) players as having single (multiple) votes. The aggregate relative power of big players is measured w.r.t.~their votes proportion. For this ratio, we show small constant worst-case bounds for the Shapley-Shubik and the Deegan-Packel indices. In sharp contrast, this ratio is unbounded for the Banzhaf index. As an application, we define a false-name strategic normal form game where each big player may split its votes between false identities, and study its various properties. Together, our results provide foundations for the implications of players' size, modeled as their ability to split, on their relative power.

Domestic “vaccine passports” are being implemented across the world as a way of increasing vaccinated people’s freedom of movement and to encourage vaccination. However, these vaccine passports may affect people’s vaccination decisions in unintended and undesirable ways. This cross-sectional study investigated whether people’s willingness and motivation to get vaccinated relate to their psychological needs (autonomy, competence and relatedness), and how vaccine passports might affect these needs. Across two countries and 1358 participants, we found that need frustration—particularly autonomy frustration—was associated with lower willingness to get vaccinated and with a shift from self-determined to external motivation. In Israel (a country with vaccine passports), people reported greater autonomy frustration than in the UK (a country without vaccine passports). Our findings suggest that control measures, such as domestic vaccine passports, may have detrimental effects on people’s autonomy, motivation, and willingness to get vaccinated. Policies should strive to achieve a highly vaccinated population by supporting individuals’ autonomous motivation to get vaccinated and using messages of autonomy and relatedness, rather than applying pressure and external controls.

We introduce the class of virtually additive non-Bayesian learning heuristics to aggregating beliefs in social networks. A virtually additive heuristic is characterized by a single function that maps a belief to a real number that represents the virtual belief. To aggregate beliefs, an agent simply sums up all the virtual beliefs of his neighbors to obtain his new virtual belief. This class of heuristics determines whether robust learning, by any naive heuristic, is possible. That is, we show that in a canonical setting with a binary state and conditionally i.i.d. signals whenever it is possible to naively learn the state robustly it is also possible to do so with a virtually additive heuristic. Moreover, we show that naive learning with virtually additive heuristics can hold without the common prior assumption.

Can we construct a neural model that is inductively biased towards learning human languages? Motivated by this question, we aim at constructing an informative prior over neural weights, in order to adapt quickly to held-out languages in the task of character-level language modeling. We infer this distribution from a sample of typologically diverse training languages via Laplace approximation. The use of such a prior outperforms baseline models with an uninformative prior (so-called "fine-tuning") in both zero-shot and few-shot settings. This shows that the prior is imbued with universal phonological knowledge. Moreover, we harness additional language-specific side information as distant supervision for held-out languages. Specifically, we condition language models on features from typological databases, by concatenating them to hidden states or generating weights with hyper-networks. These features appear beneficial in the few-shot setting, but not in the zero-shot setting. Since the paucity of digital texts affects the majority of the world's languages, we hope that these findings will help broaden the scope of applications for language technology.

Beauty and wellness are terms used often in common parlance, however their meaning and relation to each other is unclear. To probe their meaning, we applied network science methods to estimate and compare the semantic networks associated with beauty and wellness in different age generation cohorts (Generation Z, Millennials, Generation X, and Baby Boomers) and in women and men. These mappings were achieved by estimating group-based semantic networks from free association responses to a list of 47 words, either related to Beauty, Wellness, or Beauty + Wellness. Beauty was consistently related to Elegance, Feminine, Gorgeous, Lovely, Sexy, and Stylish. Wellness was consistently related Aerobics, Fitness, Health, Holistic, Lifestyle, Medical, Nutrition, and Thrive. In addition, older cohorts had semantic networks that were less connected and more segregated from each other. Finally, we found that women compared to men had more segregated and organized concepts of Beauty and Wellness. In contemporary societies that are pre-occupied by the pursuit of beauty and a healthy lifestyle, our findings shed novel light on how people think about beauty and wellness and how they are related across different age generations and by sex.

In the present study we devised a novel coding scheme for responses generated in a divergent thinking task. Based on considerations from behavioural and neurocognitive research from an embodied perspective, our scheme aims to capture dimensions of simulations of action or the body. In an exploratory investigation, we applied our novel coding scheme to analyze responses from a previously published dataset of divergent thinking responses. We show that a) these dimensions are reliably coded by naive raters, and that b) individual differences in creativity influences the way in which different dimensions are used over time. Overall, our results provide new hypotheses about the generation of creative response in the divergent thinking task and should serve to characterize the cognitive strategies used in creative endeavors.

We consider variants of the classical Frank-Wolfe algorithm for constrained smooth convex minimization, that instead of access to the standard oracle for minimizing a linear function over the feasible set, have access to an oracle that can find an extreme point of the feasible set that is closest in Euclidean distance to a given vector. We first show that for many feasible sets of interest, such an oracle can be implemented with the same complexity as the standard linear optimization oracle. We then show that with such an oracle we can design new Frank-Wolfe variants which enjoy significantly improved complexity bounds in case the set of optimal solutions lies in the convex hull of a subset of extreme points with small diameter (e.g., a low-dimensional face of a polytope). In particular, for many 0--1 polytopes, under quadratic growth and strict complementarity conditions, we obtain the first linearly convergent variant with rate that depends only on the dimension of the optimal face and not on the ambient dimension.

Few-shot crosslingual transfer has been shown to outperform its zero-shot counterpart with pretrained encoders like multilingual BERT. Despite its growing popularity, little to no attention has been paid to standardizing and analyzing the design of few-shot experiments. In this work, we highlight a fundamental risk posed by this shortcoming, illustrating that the model exhibits a high degree of sensitivity to the selection of few shots. We conduct a large-scale experimental study on 40 sets of sampled few shots for six diverse NLP tasks across up to 40 languages. We provide an analysis of success and failure cases of few-shot transfer, which highlights the role of lexical features. Additionally, we show that a straightforward full model finetuning approach is quite effective for few-shot transfer, outperforming several state-of-the-art few-shot approaches. As a step towards standardizing few-shot crosslingual experimental designs, we make our sampled few shots publicly available.

Deep learning algorithms have shown promising results in visual question answering (VQA) tasks, but a more careful look reveals that they often do not understand the rich signal they are being fed with. To understand and better measure the generalization capabilities of VQA systems, we look at their robustness to counterfactually augmented data. Our proposed augmentations are designed to make a focused intervention on a specific property of the question such that the answer changes. Using these augmentations, we propose a new robustness measure, Robustness to Augmented Data (RAD), which measures the consistency of model predictions between original and augmented examples. Through extensive experimentation, we show that RAD, unlike classical accuracy measures, can quantify when state-of-the-art systems are not robust to counterfactuals. We find substantial failure cases which reveal that current VQA systems are still brittle. Finally, we connect between robustness and generalization, demonstrating the predictive power of RAD for performance on unseen augmentations.

Schema matching is at the heart of integrating structured and semi-structured data with applications in data warehousing, data analysis recommendations, Web table matching, etc. Schema matching is known as an uncertain process and a common method to overcome this uncertainty introduces a human expert with a ranked list of possible schema matches to choose from, known as top-K K matching . In this work we propose a learning algorithm that utilizes an innovative set of features to rerank a list of schema matches and improves upon the ranking of the best match. We provide a bound on the size of an initial match list, tying the number of matches with a desired level of confidence in finding the best match. We also propose the use of matching predictors as features in a learning task, and tailored nine new matching predictors for this purpose. The proposed algorithm assists the matching process by introducing a quality set of alternative matches to a human expert. It also serves as a step towards eliminating the involvement of human experts as decision makers in a matching process altogether. A large scale empirical evaluation with real-world benchmark shows the effectiveness of the proposed algorithmic solution.

Creative cognition relies on the ability to form remote associations between concepts, which allows to generate novel ideas or solve new problems. Such an ability is related to the organization of semantic memory; yet whether real-life creative behavior relies on semantic memory organization and its neural substrates remains unclear. Therefore, this study explored associations between brain functional connectivity patterns, network properties of individual semantic memory, and real-life creativity. We acquired multi-echo functional MRI data while participants underwent a semantic relatedness judgment task. These ratings were used to estimate their individual semantic memory networks, whose properties significantly predicted their real-life creativity. Using a connectome-based predictive modeling approach, we identified patterns of task-based functional connectivity that predicted creativity-related semantic memory network properties. Furthermore, these properties mediated the relationship between functional connectivity and real-life creativity. These results provide new insights into how brain connectivity supports the associative mechanisms of creativity.

Creativity has been separately related to differences in foraging search styles and semantic memory structure. Here, we converge computational methods to examine the relation of creative foraging styles, semantic memory structure, and creative thinking. A large sample of participants was divided into groups based on their exploration and exploitation strategies in a novel creative foraging game. Their semantic memory networks were estimated and compared, based on an animal category semantic fluency task. We find differential relations between the properties of semantic memory structure and foraging styles and link such differences to performance in a standard creative thinking task. Our results highlight the interaction of semantic memory structure and foraging strategies in creativity.

We study competition among contests in a general model that allows for an arbitrary and heterogeneous space of contest design, where the goal of the contest designers is to maximize the contestants' sum of efforts. Our main result shows that optimal contests in the monopolistic setting (i.e., those that maximize the sum of efforts in a model with a single contest) form an equilibrium in the model with competition among contests. Under a very natural assumption these contests are in fact dominant, and the equilibria that they form are unique. Moreover, equilibria with the optimal contests are Pareto-optimal even in cases where other equilibria emerge. In many natural cases, they also maximize the social welfare.

Choice prediction competitions suggest that popular models of choice, including prospect theory, have low predictive accuracy. Peterson et al. show the key problem lies in assuming each alternative is evaluated in isolation, independently of the context. This observation demonstrates how a focus on predictions can promote understanding of cognitive processes.

We study linear contextual bandits with access to a large, confounded, offline dataset that was sampled from some fixed policy. We show that this problem is closely related to a variant of the bandit problem with side information. We construct a linear bandit algorithm that takes advantage of the projected information, and prove regret bounds. Our results demonstrate the ability to take advantage of confounded offline data. Particularly, we prove regret bounds that improve current bounds by a factor related to the visible dimensionality of the contexts in the data. Our results indicate that confounded offline data can significantly improve online learning algorithms. Finally, we demonstrate various characteristics of our approach through synthetic simulations.

Direct loss minimization is a popular approach for learning predictors over structured label spaces. This approach is computationally appealing as it replaces integration with optimization and allows to propagate gradients in a deep net using loss-perturbed prediction. Recently, this technique was extended to generative models, by introducing a randomized predictor that samples a structure from a randomly perturbed score function. In this work, we interpolate between these techniques by learning the variance of randomized structured predictors as well as their mean, in order to balance between the learned score function and the randomized noise. We demonstrate empirically the effectiveness of learning this balance in structured discrete spaces.

This paper develops a methodology for regret minimization with stochastic first-order oracle feedback in online, constrained, non-smooth, non-convex problems. In this setting, the minimization of external regret is beyond reach for first-order methods, and there are no gradient-based algorithmic frameworks capable of providing a solution. On that account, we propose a conceptual approach that leverages non-convex optimality measures, leading to a suitable generalization of the learner’s local regret. We focus on a local regret measure defined via a proximal-gradient mapping, that also encompasses the original notion proposed by Hazan et al. (2017). To achieve no local regret in this setting, we develop a proximal-gradient method based on stochastic first-order feedback, and a simpler method for when access to a perfect first-order oracle is possible. Both methods are order-optimal (in the min-max sense), and we also establish a bound on the number of proximal-gradient queries these methods require. As an important application of our results, we also obtain a link between online and offline non-convex stochastic optimization manifested as a new proximal-gradient scheme with complexity guarantees matching those obtained via variance reduction techniques.

We study the problem of learning conditional average treatment effects (CATE) from high-dimensional, observational data with unobserved confounders. Unobserved confounders introduce ignorance -- a level of unidentifiability -- about an individual's response to treatment by inducing bias in CATE estimates. We present a new parametric interval estimator suited for high-dimensional data, that estimates a range of possible CATE values when given a predefined bound on the level of hidden confounding. Further, previous interval estimators do not account for ignorance about the CATE associated with samples that may be underrepresented in the original study, or samples that violate the overlap assumption. Our interval estimator also incorporates model uncertainty so that practitioners can be made aware of out-of-distribution data. We prove that our estimator converges to tight bounds on CATE when there may be unobserved confounding, and assess it using semi-synthetic, high-dimensional datasets.

We propose to analyse the conditional distributional treatment effect (CoDiTE), which, in contrast to the more common conditional average treatment effect (CATE), is designed to encode a treatment's distributional aspects beyond the mean. We first introduce a formal definition of the CoDiTE associated with a distance function between probability measures. Then we discuss the CoDiTE associated with the maximum mean discrepancy via kernel conditional mean embeddings, which, coupled with a hypothesis test, tells us whether there is any conditional distributional effect of the treatment. Finally, we investigate what kind of conditional distributional effect the treatment has, both in an exploratory manner via the conditional witness function, and in a quantitative manner via U-statistic regression, generalising the CATE to higher-order moments. Experiments on synthetic, semi-synthetic and real datasets demonstrate the merits of our approach.

We consider the $[0,1]$-valued solution $(u_{t,x}:t\geq 0, x\in \mathbb R)$ to the one dimensional stochastic reaction diffusion equation with Wright-Fisher noise \[ \partial_t u = \partial_x^2 u + f(u) + \epsilon \sqrt{u(1-u)} \dot W. \] Here, $W$ is a space-time white noise, $\epsilon > 0$ is the noise strength, and $f$ is a continuous function on $[0,1]$ satisfying $\sup_{z\in [0,1]}|f(z)|/ \sqrt{z(1-z)} < \infty.$ We assume the initial data satisfies $1 - u_{0,-x} = u_{0,x} = 0$ for $x$ large enough. Recently, it was proved in (Comm. Math. Phys. 384 (2021), no. 2) that the front of $u_t$ propagates with a finite deterministic speed $V_{f,\epsilon}$, and under slightly stronger conditions on $f$, the asymptotic behavior of $V_{f,\epsilon}$ was derived as the noise strength $\epsilon$ approaches $\infty$. In this paper we complement the above result by obtaining the asymptotic behavior of $V_{f,\epsilon}$ as the noise strength $\epsilon$ approaches $0$: for a given $p\in [1/2,1)$, if $f(z)$ is non-negative and is comparable to $z^p$ for sufficiently small $z$, then $V_{f,\epsilon}$ is comparable to $\epsilon^{-2\frac{1-p}{1+p}}$ for sufficiently small $\epsilon$.

Bayes-rational agents reside on a social network. They take binary actions sequentially and irrevocably, and the right action depends on an unobservable state. Each agent receives a bounded private signal about the realized state and observes the actions taken by the neighbors who acted before. How does the network topology affect the ability of agents to aggregate the information dispersed over the population by means of the private signals? Most of the literature addressing such questions assumes that the network's structure is dictated by the order in which agents take their actions. By contrast, we assume that the network preexists and the order in which agents take actions is random. Hence, the network's topology is decoupled from the order of actions in a particular decision problem. The random order leads to a novel localization phenomenon: for most of the orders, agents have a bounded radius of influence, i.e., the agent's action is unlikely to affect those who are far from him in the network. This phenomenon underlies a bunch of new effects. Global information cascades become unlikely, and networks that fail to aggregate information exhibit many local cascades. The ability of an agent to learn the right action is determined by the local structure of the network around him, and there is a local topological condition guaranteeing that the agent takes the right action no matter how well others do. Roughly speaking, the condition requires that the agent bridges a multitude of mutually exclusive social circles. Networks, where this condition is satisfied for all agents, are robust to disruptions and keep aggregating information even if a substantial fraction of the population is eliminated adversarially.

Through a sequence of examples, we survey the main results of "Feasible Joint Posterior Beliefs" [Arieli, Babichenko, Sandomirskiy, Tamuz 2021]. A group of agents share a common prior distribution regarding a binary state, and observe some information structure. What are the possible joint distributions of their posteriors? We discuss feasibility of product distributions, correlation of posteriors in feasible distributions, extreme feasible distributions and the characterization of feasibility in terms of a "no-trade" condition.

An essential aspect of human communication is the ability to access and retrieve information from ones’ ‘mental lexicon’. This lexical access activates phonological and semantic components of concepts, yet the question whether and how these two components relate to each other remains widely debated. We harness tools from network science to construct a large-scale linguistic multilayer network comprising of phonological and semantic layers. We find that the links in the two layers are highly similar to each other and that adding information from one layer to the other increases efficiency by decreasing the network overall distances, but specifically affecting shorter distances. Finally, we show how a multilayer architecture demonstrates the highest efficiency, and how this efficiency relates to weak semantic relations between cue words in the network. Thus, investigating the interaction between the layers and the unique benefit of a linguistic multilayer architecture allows us to quantify theoretical cognitive models of lexical access.

A population of voters must elect representatives among themselves to decide on a sequence of possibly unforeseen binary issues. Voters care only about the final decision, not the elected representatives. The disutility of a voter is proportional to the fraction of issues, where his preferences disagree with the decision. While an issue-by-issue vote by all voters would maximize social welfare, we are interested in how well the preferences of the population can be approximated by a small committee. We show that a k-sortition (a random committee of k voters with the majority vote within the committee) leads to an outcome within the factor 1+O(1/√ k) of the optimal social cost for any number of voters n, any number of issues m, and any preference profile. For a small number of issues m, the social cost can be made even closer to optimal by delegation procedures that weigh committee members according to their number of followers. However, for large m, we demonstrate that the k-sortition is the worst-case optimal rule within a broad family of committee-based rules that take into account metric information about the preference profile of the whole population.