Results 11 to 20 of about 277,892 (265)
Editorial: Safe and Trustworthy Machine Learning [PDF]
Frontiers in Big Data, 2021 Bhavya Kailkhura +3 more
doaj +3 more sources
Learning to be safe, in finite time [PDF]
2021 American Control Conference (ACC), 2021 This paper aims to put forward the concept that learning to take safe actions in unknown environments, even with probability one guarantees, can be achieved without the need for an unbounded number of exploratory trials, provided that one is willing to relax its optimality requirements mildly.
Agustin Castellano +2 more
openaire +3 more sources
Learning to Recover for Safe Reinforcement Learning
CoRR, 2023 Safety controllers is widely used to achieve safe reinforcement learning. Most methods that apply a safety controller are using handcrafted safety constraints to construct the safety controller. However, when the environment dynamics are sophisticated, handcrafted safety constraints become unavailable.
Haoyu Wang, Xin Yuan, Qinqing Ren
openaire +3 more sources
Safe Weakly Supervised Learning [PDF]
Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, 2021 Weakly supervised learning (WSL) refers to learning from a large amount of weak supervision data. This includes i) incomplete supervision (e.g., semi-supervised learning); ii) inexact supervision (e.g., multi-instance learning) and iii) inaccurate supervision (e.g., label noise learning). Unlike supervised learning which typically achieves performance
openaire +2 more sources
Safe Distributional Reinforcement Learning [PDF]
, 2022 Safety in reinforcement learning (RL) is a key property in both training and execution in many domains such as autonomous driving or finance. In this paper, we formalize it with a constrained RL formulation in the distributional RL setting. Our general model accepts various definitions of safety(e.g., bounds on expected performance, CVaR, variance, or ...
Jianyi Zhang, Paul Weng
openaire +4 more sources
Safe Learning for Near-Optimal Scheduling [PDF]
, 2021 In this paper, we investigate the combination of synthesis, model-based learning, and online sampling techniques to obtain safe and near-optimal schedulers for a preemptible task scheduling problem. Our algorithms can handle Markov decision processes (MDPs) that have 1020 states and beyond which cannot be handled with state-of-the art probabilistic ...
Damien Busatto-Gaston +4 more
openaire +6 more sources
A Computationally Lightweight Safe Learning Algorithm
2023 62nd IEEE Conference on Decision and Control (CDC), 2023 Safety is an essential asset when learning control policies for physical systems, as violating safety constraints during training can lead to expensive hardware damage. In response to this need, the field of safe learning has emerged with algorithms that can provide probabilistic safety guarantees without knowledge of the underlying system dynamics ...
Dominik Baumann +3 more
openaire +4 more sources
Curriculum learning for safe mapless navigation
Proceedings of the 37th ACM/SIGAPP Symposium on Applied Computing, 2022 8 pages, 5 figures. The poster version of this paper has been accepted by The 37th ACM/SIGAPP Symposium on Applied Computing Proceedings (SAC IRMAS 2022)
Marzari, Luca +3 more
openaire +4 more sources
Virtuously Safe Reinforcement Learning
CoRR, 2018 We show that when a third party, the adversary, steps into the two-party setting (agent and operator) of safely interruptible reinforcement learning, a trade-off has to be made between the probability of following the optimal policy in the limit, and the probability of escaping a dangerous situation created by the adversary.
C. Henrik Åslund +3 more
openaire +3 more sources
Dual Learning-Based Safe Semi-Supervised Learning
IEEE Access, 2018 In many real-world applications, labeled instances are generally limited and expensively collected, while the most instances are unlabeled and the amount is often sufficient.
Haitao Gan +3 more
doaj +1 more source