skip to main content


Title: Risk-Sensitive Reinforcement Learning: Near-Optimal Risk-Sample Tradeoff in Regret
Award ID(s):
1955997
NSF-PAR ID:
10249847
Author(s) / Creator(s):
; ; ; ;
Date Published:
Journal Name:
Advances in neural information processing systems
Volume:
33
ISSN:
1049-5258
Page Range / eLocation ID:
22384--22395
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Testing multiple subjects within a group, with a single test applied to the group (i.e., group testing), is an important tool for classifying populations as positive or negative for a specific binary characteristic in an efficient manner. We study the design of easily implementable, static group testing schemes that take into account operational constraints, heterogeneous populations, and uncertainty in subject risk, while considering classification accuracy- and robustness-based objectives. We derive key structural properties of optimal risk-based designs and show that the problem can be formulated as network flow problems. Our reformulation involves computationally expensive high-dimensional integrals. We develop an analytical expression that eliminates the need to compute high-dimensional integrals, drastically improving the tractability of constructing the underlying network. We demonstrate the impact through a case study on chlamydia screening, which leads to the following insights: (1) Risk-based designs are shown to be less expensive, more accurate, and more robust than current practices. (2) The performance of static risk-based schemes comprised of only two group sizes is comparable to those comprised of many group sizes. (3) Static risk-based schemes are an effective alternative to more complicated dynamic schemes. (4) An expectation-based formulation captures almost all benefits of a static risk-based scheme. 
    more » « less
  2. The majority of human-factor models in construction safety assume that risk-taking behaviors, failure to perceive hazards, or misinterpreting the associated risks of hazards are the main contributing factors in accident occurrences. However, the findings for the link between risk-taking behaviors and risk perception are inconsistent. To address this knowledge gap, the current study focuses on measuring the association between risk perception and the risk-taking behaviors of construction workers. To achieve this objective, 27 undergraduate students from the University of Nebraska–Lincoln with at least 1 year of experience in the construction industry were recruited to participate in an experiment. To measure risk perception, the subjects were asked to assess the risk—in terms of likelihood and severity—associated with various scenario statements related to fall hazards. Subsequently, subjects performed the balloon analogue risk task (BART), a computerized decision-making simulation, to test the subjects’ risk-taking behaviors. The results of a correlational analysis showed that there is a significant negative association between an individual’s risk perception of fall hazards and his/her risk-taking behaviors. Additionally, differences in the risk-taking behaviors of subjects evaluated against their risk-perception scores were examined using a permutation simulation analysis. The results showed that there is a moderately significant difference in the risk-taking behaviors of subjects with low and high fall-risk perception. The research findings provide empirical evidence that people with lower risk perception tend to engage in more risk-taking behaviors. Furthermore, this study is one of the first attempts at using BART in the assessment of risk taking in construction safety and paves the way for a better understanding the human factors that contribute to construction accidents. 
    more » « less