Search · four archives
Search · four archives
24 papers · ranked by Valyu relevance
Shangding Gu, Long Yang, Yali Du, Guang Chen + 4 more
'Jun Wang' 'Yaodong Yang' 'Alois Knoll'] - A review of safe Reinforcement Learning (RL) methods is provided with theoretical analysis and application analysis. - The key question that safe RL needs to answer is proposed, and five problems "2H3W" are analysed to address the key question. - To examine the effectiveness…
Mahmoud Selim, Amr Alanwar, M. Watheq El‐Kharashi, Hazem M. Abbas + 1 more
'Karl Henrik Johansson'] Abstract—Reinforcement learning (RL) algorithms can achieve state-of-the-art performance in decision-making and continuous control tasks. However, applying RL algorithms on safety-critical systems still needs to be well justified due to the exploration nature of many RL algorithms, especially…
Nopparat Songserm, Rapeepan Pitakaso, Thanatkij Srichok, Surajet Khonjun + 7 more
Highlights Public health relevance-How does this work relate to a public health issue?1. The proposed multi-agent framework supports faster and more adaptive decision-making in complex public service environments. 2. The system can help improve coordination and resource management in situations that require continuous…
Krishnan Srinivasan, Benjamin Eysenbach, Sehoon Ha, Jie Tan + 1 more
'Chelsea Finn'] Safety is an essential component for deploying reinforcement learning (RL) algorithms in real-world scenarios, and is critical during the learning process itself. A natural first approach toward safe RL is to manually specify constraints on the policy's behavior. However, just as learning has enabled…
Oluwatosin Oseni, Shengjie Wang, Jun Zhu, Micah Corah
—Reinforcement Learning (RL) has shown remarkable success in real-world applications, particularly in robotics control. However, RL adoption remains limited due to insufficient safety guarantees. We introduce Nightmare Dreamer, a modelbased Safe RL algorithm that addresses safety concerns by leveraging a learned world…
Rui Zhao, Ziguo Chen, Yuze Fan, Yun Li + 2 more
Reinforcement Learning (RL) methods are regarded as effective for designing autonomous driving policies. However, even when RL policies are trained to convergence, ensuring their robust safety remains a challenge, particularly in long-tail data. Therefore, decision-making based on RL must adequately consider potential…
Daniel Bethell, Simos Gerasimou, Radu Călinescu, Calum Imrie
Ensuring the safe exploration of reinforcement learning (RL) agents is critical for deployment in real-world systems. Yet existing approaches struggle to strike the right balance: methods that tightly enforce safety often cripple task performance, while those that prioritise reward leave safety constraints frequently…
De-Tian Chu, Lin-Yuan Bai, Jia-Nuo Huang, Zhen-Long Fang + 4 more
'Peng Zhang' 'Wei Kang' 'Hai-Feng Ling' 'Wenling Li'] Ensuring safety in autonomous driving is crucial for effective motion planning and navigation. However, most end-to-end planning methodologies lack sufficient safety measures. This study tackles this issue by formulating the control optimization problem in…
Mohammad Ali Labbaf Khaniki, Morteza Mirzaee, Elahe Moradi
Non-stationary environments pose significant challenges for reinforcement learning (RL), particularly in safety-critical applications like robotics and energy systems, where adaptability, stability, and robustness to uncertainty are essential. This paper introduces Lyapunov-Guided Distributional Reinforcement Learning…
Mingyu Cai, Shaoping Xiao, Junchao Li, Zhen Kan
This paper proposes an advanced Reinforcement Learning (RL) method, incorporating reward-shaping, safety value functions, and a quantum action selection algorithm. The method is model-free and can synthesize a finite policy that maximizes the probability of satisfying a complex task. Although RL is a promising…
Homayoun Honari, Mehran Ghafarian Tamizi, Homayoun Najjaran
— Safe reinforcement learning (Safe RL) refers to a class of techniques that aim to prevent RL algorithms from violating constraints in the process of decision-making and exploration during trial and error. In this paper, a novel modelfree Safe RL algorithm, formulated based on the multi-objective policy optimization…
Shangding Gu, Alap Kshirsagar, Yali Du, Guang Chen + 2 more
'Alois Knoll'] Deployment of Reinforcement Learning (RL) algorithms for robotics applications in the real world requires ensuring the safety of the robot and its environment. Safe Robot RL (SRRL) is a crucial step toward achieving human-robot coexistence. In this paper, we envision a human-centered SRRL framework…
André Correia, Luı́s A. Alexandre
Deploying reinforcement learning agents in the real world can be challenging due to the risks associated with learning through trial and error. We propose a task-agnostic method that leverages small sets of safe and unsafe demonstrations to improve the safety of RL agents during learning. The method compares the…
Authors not listed
Realizing the promise of artificial intelligence (AI) to accelerate scientific progress and deliver technological impact depends on how effectively AI can be integrated into real-world decision- making processes. As Peter Norvig states, “Somewhat remarkably, almost all AI research until very recently has assumed that…
Pranav Mahajan, Shuangyi Tong, Sang Wan Lee, Ben Seymour
The safety-efficiency dilemma describes the problem of maintaining safety during efficient exploration, and is a special case of the exploration-exploitation dilemma in the face of potentially catastrophic dangers. Conventional exploration-exploitation solutions collapse punishment and reward into a single signal…
Pranav Mahajan, Ben Seymour
The seminal reward prediction error theory of dopamine function faces several key challenges. Most notable is the difficulty learning multiple rewards simultaneously, inefficient on-policy learning, and accounting for heterogeneous striatal responses in the tail of the striatum. We propose a normative framework, based…
Felix Berkenkamp, Andreas Krause, Angela P. Schoellig
Selecting the right tuning parameters for algorithms is a pravelent problem in machine learning that can significantly affect the performance of algorithms. Data-efficient optimization algorithms, such as Bayesian optimization, have been used to automate this process. During experiments on real-world systems such as…
Yoshihisa Fujita, Sho Yagishita, Haruo Kasai, Shin Ishii
Generalization enables applying past experience to similar but nonidentical situations. Therefore, it may be essential for adaptive behaviors. Recent neurobiological observation indicates that the striatal dopamine system achieves generalization and subsequent discrimination by updating corticostriatal synaptic…
Dharanish Rajendra, Chaitanya S. Gokhale
Animals adapt their behaviour to current environmental conditions to enhance survival and reproductive success. While longterm adaptation occurs through evolutionary processes acting on heritable variation, individuals can also adapt within their lifetime via learning. Learning is particularly advantageous in…
Pranav Mahajan, Ben Seymour
The seminal reward prediction error account of dopamine has been highly successful, but faces several key challenges. Most notable are the difficulty of learning multiple rewards simultaneously, inefficient on-policy learning, and accounting for the heterogeneous striatal responses observed across and within striatal…
Authors not listed
Machine olfaction—the artificial replication of the sense of smell—faces significant challenges due to the absence of large, standardized training datasets. Unlike vision, language, and audio models, which benefit from extensive corpora such as ImageNet, GLUE, and AudioSet, olfaction lacks scaled equivalents and…
Feng Feng, Zhenru Chen, Jianyuan Ni, Yuanxun Zhang + 3 more
Drinking water is essential to public health and socioeconomic growth. Therefore, assessing and ensuring drinking water supply is a critical task in modern society. Conventional approaches to analyzing and controlling drinking water quality are labor-intensive and costly with a low throughput. Machine learning (ML) is…
Authors not listed
Computational toxicology plays a pivotal role in modern drug discovery and environmental risk assessment; however, the reliability of predictive models on unseen chemical scaffolds remains a critical bottleneck. Deep learning architectures, despite their prevalence, are susceptible to ’silent failures’—yielding…
Camilla Kao, Che-I Kao, Russell Furr
In science, safety can seem unfashionable. Satisfying safety requirements can slow the pace of research, make it cumbersome, or cost significant amounts of money. The logic of rules can seem unclear. Compliance can feel like a negative incentive. So besides the obvious benefit that safety keeps one safe, why do some…