Search Results (33)

Zero-shot generalization to out-of-distribution (OOD) teammates and opponents in multi-agent systems (MASs) remains a fundamental challenge for general-purpose AI, especially in open-ended interaction scenarios. Existing multi-agent reinforcement learning (MARL) paradigms, such as self-play and population-based training, often collapse to a limited subset of Nash equilibria, leaving agents brittle when faced with semantically diverse, unseen behaviors. Recent approaches that invoke Large Language Models (LLMs) at run time can improve adaptability but introduce substantial latency and can become less reliable as task horizons grow; in contrast, LLM-assisted reward-shaping methods remain constrained by the inefficiency of the inner reinforcement-learning loop. To address these limitations, we propose LLM-TOC (LLM-Driven Theory-of-Mind Adversarial Curriculum), which casts generalization as a bi-level Stackelberg game: in the inner loop, a MARL agent (the follower) minimizes regret against a fixed population, while in the outer loop, an LLM serves as a semantic oracle that generates executable adversarial or cooperative strategies in a Turing-complete code space to maximize the agent’s regret. To cope with the absence of gradients in discrete code generation, we introduce Gradient Saliency Feedback, which transforms pixel-level value fluctuations into semantically meaningful causal cues to steer the LLM toward targeted strategy synthesis. We further provide motivating theoretical analysis via the PAC-Bayes framework, showing that LLM-TOC converges at rate $O(1/\sqrt{K})$ and yields a tighter generalization error bound than parameter-space exploration under reasonable preconditions. Experiments on the Melting Pot benchmark demonstrate that, with expected cumulative collective return as the core zero-shot generalization metric, LLM-TOC consistently outperforms self-play baselines (IPPO and MAPPO) and the LLM-inference method Hypothetical Minds across all held-out test scenarios, reaching 75% to 85% of the upper-bound performance of Oracle PPO. Meanwhile, with the number of RL environment interaction steps to reach the target relative performance as the core efficiency metric, our framework reduces the total training computational cost by more than 60% compared with mainstream baselines. Full article

As people spend more time indoors, the impact of the built environment on psychological health has attracted growing attention. Yet existing studies often have difficulty capturing decision-makers’ reference dependence and loss aversion under uncertainty. To bridge this gap, we propose an evaluation framework comprising three first-level criteria—Outdoor Environment, Physical Comfort (including thermal, lighting, and color environments), and Acoustic Comfort—and determine combined weights by integrating subjective analytic hierarchy process (AHP) judgments with objective entropy weighting based on triangular fuzzy numbers (TFNs). We further incorporate prospect–regret theory to represent loss aversion, expectation-based reference points, and counterfactual regret/rejoicing, and couple it with the VIKOR compromise ranking method, forming an integrated “TFN + Prospect–Regret + VIKOR” approach. The proposed method is applied to four retrofit alternatives for an open-plan office floor (approximately 1200 m²), each emphasizing outdoor environment, physical comfort, acoustic comfort, or no single priority. Experts assessed the schemes using fuzzy linguistic variables. The results show that lighting conditions, thermal comfort, color scheme, and internal noise control receive the highest comprehensive weights. Extensive sensitivity analyses across value/weighting functions and regret-aversion parameters indicate that the ranking of alternatives remains stable while exhibiting clearer separation. Comparative analyses further suggest that, although the overall ordering is consistent with baseline methods, the proposed model increases score dispersion and improves discriminative power. Overall, by explicitly accounting for decision-makers’ psychological behavior and information uncertainty, the framework enables robust and interpretable selection of retrofit schemes for existing office spaces. Full article

We introduce Causal Incentive Design (CID), a framework that applies causal inference to canonical single-stage principal–agent problems (PAPs) characterized by bilateral private information. Within CID, the operating rules of PAPs are formalized using an additive-noise causal graphical model (CGM). Incentives are modeled as interventions on a function space variable, $\Gamma$, which correspond to policy interventions in the principal–follower causal relation. The causal inference target estimand $V(\Gamma )$ is defined as the expected value of the principal’s utility variable under a specified policy intervention in the post-intervention distribution. In the context of additive-Gaussian independent noise, the estimand $V(\Gamma )$ decomposes into a two-layer expectation: (i) an inner Gaussian smoothing of the principal’s utility regression; and (ii) an outer averaging over the conditional probability of the follower’s action given the incentive policy. A Gauss–Hermite quadrature method is employed to efficiently estimate the first layer, while a policy-local kernel reweighting approach is used for the second. For offline selection of a single incentive policy, a Functional Causal Bayesian Optimization (FCBO) algorithm is introduced. This algorithm models the objective functional $\gamma \mapsto V\left(\gamma \right)$ using a functional Gaussian process surrogate defined on a Reproducing Kernel Hilbert Space (RKHS) domain and utilizes an Upper Confidence Bound (UCB) acquisition functional. Consequently, the policy value $V\left(\gamma \right)$ becomes an interventional query that can be answered using offline observational data under standard identifiability assumptions. High-probability cumulative-regret bounds are established in terms of differential information gain for the proposed FBO algorithm. Collectively, these elements constitute the central contributions of the CID framework, which integrates causal inference through identification and estimation with policy search in principal–agent problems under private information. This approach establishes a causal decision-making pipeline that enables commitment to a high-performing incentive in a single-shot game, supported by regret guarantees. Provided that the data used for estimation is sufficient, the resulting offline pipeline is appropriate for scenarios where adaptive deployment is impractical or costly. Beyond the methodological contribution, this work introduces a novel application of causal graphical models and causal reasoning to incentive design and principal–agent problems, which are central to economics and multi-agent systems. Full article

Decision-making theory has developed over many decades at the intersection of economics, mathematics, psychology, and engineering. Its classical foundations include Bernoulli’s expected utility theory, von Neumann and Morgenstern’s rational choice theory, and the criteria proposed by Savage, Wald, Hurwicz, and others. However, in real-world contexts, decisions are made under uncertainty, incompleteness, and unreliability of information, which classical approaches do not adequately address. To overcome these limitations, modern multi-criteria decision-making methods such as Analytic Hierarchy Process (AHP), Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), VlseKriterijumska Optimizacija I Kompromisno Resenje (Compromise solution approach) (VIKOR), and ELimination Et Choix Traduisant la REalité (Elimination and Choice Expressing Reality) (ELECTRE), as well as their fuzzy and Z-number extensions, are widely applied to the modeling and evaluation of complex systems. These Z-number extensions are based on the concept of Z-numbers introduced by Lotfi Zadeh in 2011 to formalize higher-order uncertainty. This study introduces the Z-Regret principle, which extends Savage’s regret criterion through the use of Z-numbers. Supported by Rafik Aliev’s mathematical justifications concerning arithmetic operations on Z-numbers, the model evaluates regret not only as a loss relative to the best alternative but also by incorporating the degree of confidence and reliability of this evaluation. Calculations for the selection of digital advertising platforms in terms of performance assessment under various scenarios demonstrate that the Z-Regret principle enables more stable and well-founded decision-making under uncertainty. Full article

Motherhood regret still constitutes a major taboo that limits the possibility of processing the negative exposure to being a mother. This qualitative study involved Italian women living both in Italy and abroad, where traditional patriarchal thinking remains influential. Sixteen women defining themselves as ‘regretful were interviewed to explore their experiences of regret, the changes following the birth of children, family and social support, and employment. The thematic analysis highlighted several recurring themes: the idealisation of motherhood and the hidden struggles it conceals; the guilt associated with feeling inadequate and the indifference of some fathers; the social pressure that compels women to conform to maternal expectations; the perception of being trapped in a predefined role; and the conflict between personal identity and the ideal of the “perfect mother.”. The findings reveal that maternal regret is deeply intertwined with internalised patriarchal norms, the myth of the “good mother,” and the social expectation of women’s self-sacrifice. Despite recognising these as cultural constructs, participants expressed feelings of guilt, anger, and inadequacy, intensified by the unequal division of domestic and parental responsibilities. This issue and the need for a revival of women’s consciousness-raising groups to open a space for dialogue on the topic in countries where patriarchy is still strong, such as Italy, are discussed. Full article

This study investigates the influence of FOMO, loss aversion, and herd behavior on gold investment decisions in the Vietnamese market. Employing data collected from 727 investors and the Partial Least Squares Structural Equation Modeling (PLS-SEM) method, the analysis results confirm the pivotal role of FOMO, with both direct and indirect impacts on gold investment decisions. Notably, both loss aversion and herd behavior positively influence FOMO, thereby indirectly encouraging relatively hasty and inadequately considered investment decisions. The study also finds that FOMO has a negative relationship with anticipated regret but is positively correlated with subjective expected pleasure. Furthermore, as determined through Multi-Group Analysis (MGA), psychological messages featuring “self-decision” or “risk warning” demonstrate a significant moderating role, potentially reducing or enhancing the influence of FOMO on investment decisions. These findings contribute to enriching behavioral finance theory and provide an empirical basis for developing effective risk management policies and gold market regulation aimed at mitigating the negative impacts of FOMO. Full article

Introduction: Total knee arthroplasty (TKA) is a treatment for knee pain, but some patients are not satisfied with their outcomes. Utilizing shared decision making (SDM) can lead to better decisions, satisfaction, and fewer regrets. However, healthcare professionals have little knowledge of risk factors for regret. The aim of this study is to evaluate decisional regret using the Decision Regret Scale (DRS) after primary TKA among patients who engaged in SDM. Method: A total of 118 patients who underwent TKA surgery between March 2020 and May 2022 participated in this study, and they were able to reflect on their outcomes. The primary outcome was decisional regret assessed using the DRS, and the secondary outcome was post-operative pain at a three-month follow-up, measured using the Lequesne Index. Result: The study found that 49% of the patients reported no regret, 25% reported mild regret, and 26% reported moderate-to-severe regret. There was a significant correlation between greater levels of decision regret and a higher three-month Lequesne Index. Post-operative pain and post-operative mobility status and the range of motion of the knee joint were also strongly correlated. Conclusion: The study found that more than half of the patients undergoing primary TKAs experienced regret even following SDM counseling. Regret levels were associated with higher post-operative pain and poorer mobility. This underscores the importance of informing patients about potential adverse effects of TKA to manage their expectations and reduce regret in future SDM interviews. Practice implications: This study incorporated patient perspectives through their direct engagement in the SDM process prior to surgery. Patients participated in the design of the SDM framework, which included educational pamphlets and structured interviews to assess their values and preferences. Their involvement ensured that the SDM procedure was tailored to patient-centered outcomes. Furthermore, the follow-up assessments were conducted with patients to evaluate decisional regret and post-operative outcomes, providing valuable insights into the effectiveness of the SDM process. By actively participating in the research through decision making and outcome reflection, the patients contributed to the understanding of factors influencing decisional regret after undergoing TKA. Full article

In modern communication, the electromagnetic spectrum serves as the carrier for information transmission, and the only medium enabling information exchange anywhere, anytime. To adapt to the changing dynamics of a complex electromagnetic environment, electromagnetic spectrum allocation algorithms must not only meet the demands for efficiency and intelligence but also possess anti-jamming capabilities to achieve the best communication effect. Focusing on intelligent wireless communication, this paper proposes a multi-agent hybrid game spectrum allocation method under incomplete information and based on the Monte Carlo counter-factual regret minimization algorithm. Specifically, the method first utilizes frequency usage and interference information from both sides to train agents through extensive simulations using the Monte Carlo Method, allowing the trial values to approach the expected values. Based on the results of each trial, the counterfactual regret minimization algorithm is employed to update the frequency selection strategies for both the user and the interferer. Subsequently, the trained agents from both sides engage in countermeasure communication. Finally, the probabilities of successful communication and successful interference for both sides are statistically analyzed. The results show that under the multi-agent hybrid game spectrum allocation method based on the Monte Carlo counter-factual regret minimization algorithm, the probability of successful interference against the user is 32.5%, while the probability of successful interference by the jammer is 37.3%. The average simulation time per round is 3.06 s. This simulation validates the feasibility and effectiveness of the multi-agent hybrid game spectrum allocation module based on the Monte Carlo counter-factual regret minimization algorithm. Full article

Non-stationary multi-armed bandit (MAB) problems have recently attracted extensive attention. We focus on the abruptly changing scenario where reward distributions remain constant for a certain period and change at unknown time steps. Although Thompson sampling (TS) has shown success in non-stationary settings, there is currently no regret bound analysis for TS with uninformative priors. To address this, we propose two algorithms, discounted TS and sliding-window TS, designed for sub-Gaussian reward distributions. For these algorithms, we establish an upper bound for the expected regret by bounding the expected number of times a suboptimal arm is played. We show that the regret upper bounds of both algorithms are $\tilde{O}(\sqrt{T{B}_T})$, where T is the time horizon and $B_T$ is the number of breakpoints. This upper bound matches the lower bound for abruptly changing problems up to a logarithmic factor. Empirical comparisons with other non-stationary bandit algorithms highlight the competitive performance of our proposed methods. Full article

Federated learning (FL) is a distributed machine learning approach that can enable Internet of Things (IoT) edge devices to collaboratively learn a machine learning model without explicitly sharing local data in order to achieve data clustering, prediction, and classification in networks. In previous works, some online multi-armed bandit (MAB)-based FL frameworks were proposed to enable dynamic client scheduling for improving the efficiency of FL in underwater wireless IoT networks. However, the security of online dynamic scheduling, which is especially essential for underwater wireless IoT, is increasingly being questioned. In this work, we study secure dynamic scheduling for FL frameworks that can protect against malicious clients in underwater FL-assisted wireless IoT networks. Specifically, in order to jointly optimize the communication efficiency and security of FL, we employ MAB-based methods and propose upper-confidence-bound-based smart contracts (UCB-SCs) and upper-confidence-bound-based smart contracts with a security prediction model (UCB-SCPs) to address the optimal scheduling scheme over time-varying underwater channels. Then, we give the upper bounds of the expected performance regret of the UCB-SC policy and the UCB-SCP policy; these upper bounds imply that the regret of the two proposed policies grows logarithmically over communication rounds under certain conditions. Our experiment shows that the proposed UCB-SC and UCB-SCP approaches significantly improve the efficiency and security of FL frameworks in underwater wireless IoT networks. Full article

In the scenario of an integrated space–air–ground emergency communication network, users encounter the challenge of rapidly identifying the optimal network node amidst the uncertainty and stochastic fluctuations of network states. This study introduces a Multi-Armed Bandit (MAB) model and proposes an optimization algorithm leveraging dynamic variance sampling (DVS). The algorithm posits that the prior distribution of each node’s network state conforms to a normal distribution, and by constructing the distribution’s expected value and variance, it maximizes the utilization of sample data, thereby maintaining an equilibrium between data exploitation and the exploration of the unknown. Theoretical substantiation is provided to illustrate that the Bayesian regret associated with the algorithm exhibits sublinear growth. Empirical simulations corroborate that the algorithm in question outperforms traditional ε-greedy, Upper Confidence Bound (UCB), and Thompson sampling algorithms in terms of higher cumulative rewards, diminished total regret, accelerated convergence rates, and enhanced system throughput. Full article

Background: Identifying the potential factors of depression among medical students is the first step towards academic excellence and future safe medical practice. Methods: A cross-sectional study was conducted from December 2019 to February 2020 at the University of Bisha, College of Medicine (UBCOM), Bisha Province, Saudi Arabia. Male medical students from year one to year six were involved. A self-administered questionnaire was used to collect data about students’ socio-demographic and academic characteristics. The Arabic version of the PHQ-9 scale with a score of ≥10 was used to identify depression. Logistic regression analysis was used to assess the prevalence and correlates of depression. Results: Of the 190 male students enrolled, 26.8% had depressive symptoms, of whom 45.1% were experiencing moderate to severe symptoms. The significantly highest depression rate was found among the second-year students, at 43.8% (OR = 2.544; 95% CI 1.178–5.714; p = 0.018), and the lowest rate was found among year one students, at 8.9% (OR = 0.203; 95% CI 0.075–0.560; p = 0.002). Univariate regression revealed a significant correlation between depression and dissatisfaction with family income, loss of family members, having psychological illness, difficulties in personal relationships, regretting studying medicine, failure in an academic year, a lower grade than expected, conflict with tutors, lack of college facilities and heavy academic load. In multivariate analysis, loss of family members (AOR = 3.69; 95% CI 1.86–7.413), difficulties in personal relationships (AOR = 2.371; 95% CI 1.009–5.575), regretting studying medicine (AOR = 3.764; 95% CI 1.657–8.550), and failing an academic year (AOR = 2.559; 95% CI 1.112–5.887) were independently correlated with depression. Conclusions: The study concluded that medical students at UBCOM experience depressive symptoms associated with various risk indicators. Optimizing the educational and social environment and infrastructure facilities at UBCOM might promote students’ mental health and well-being. Full article

This paper critically examines the weighted sample average approximation (wSAA) framework, a widely used approach in prescriptive analytics for managing uncertain optimization problems featuring non-linear objectives. Our research pinpoints a key deficiency of the wSAA framework: when data samples are limited, the minimum relative regret—the discrepancy between the expected optimal profit realized by an oracle aware of the genuine distribution, and the maximum expected out-of-sample profit garnered by the data-driven policy, normalized by the former profit—can approach towards one. To validate this assertion, we scrutinize two distinct contextual stochastic optimization problems—the production decision-making problem and the ship maintenance optimization problem—within the wSAA framework. Our study exposes a potential deficiency of the wSAA framework: its decision performance markedly deviates from the full-information optimal solution under limited data samples. This finding offers valuable insights to both researchers and practitioners employing the wSAA framework. Full article

The theoretical literature on risk aversion and Expected Utility Theory is extensive; however, the analysis of this behaviour with natural experiments could be more comprehensive. In this paper, we use data from 120 episodes of the Portuguese version of the TV game show The Price is Right, namely from The Wheel game, to explore risk aversion as well as the impact of gender in decision-making. The Wheel game has straightforward rules and huge expected payoffs. All contestants have access to the same information and distributions of uncertainty, making it a unique field laboratory to conduct an experimental test of rational decision theory. The objective is to infer the risk aversion levels of decision-makers from their choice to turn the wheel and the influence of gender on risk attitudes. There is a widespread view that women are more risk-averse than men. However, we could not reject the hypothesis that women and men have the same level of risk aversion. Nevertheless, we have evidence that contestants are more risk-averse than risk-seeking. The omission bias, loss aversion and regret can explain that behaviour. Full article

The changing economic environment in Taiwan has led to an increase in the structure of double-income families. To compensate for the lack of time to take care of their children and the regret of their learning process, parents will send their children to kindergarten early. Parents choose to expose their children to better education and learning models because they do not want their children to get behind at the starting point. The newly introduced unplugged programming language curriculum can develop children’s logical and computational thinking skills to face future learning and employment skills in information and communications technology-related industries. The purpose of this study is to examine the parental acceptance of unplugged programming language courses and to analyze the relationship between the variables in the study framework to understand influencing factors. The theoretical basis of the study is the planned behavior theory. This study replaces behavioral intention with parents’ acceptance and establishes a basic framework of attitude, subjective norm, and perceived behavioral control. The study framework is established by combining the factors of expectation and compensation as antecedent variables of attitude. An online e-questionnaire is distributed to parents of children aged 5–6 years old in Taiwan to collect data. The structural equation model is conducted on 489 data points. Results of the study reveal that expectation and compensation have a significant effect on attitude. Attitude, subjective norm, and perceived behavioral control have a significant positive effect on family acceptance of unplugged programming language. Furthermore, the expectation and compensation psychology affect the parental acceptance of unplugged language programs through attitude. Finally, practical applications and future research directions regarding the promotion of unplugged programming language for young children are provided. Full article