Open AccessPosted Content
Trial without Error: Towards Safe Reinforcement Learning via Human Intervention
TLDR
This work formalizes human intervention for RL and shows how to reduce the human labor required by training a supervised learner to imitate the human's intervention decisions, and outlines extensions of the scheme that are necessary if the authors are to train model-free agents without a single catastrophe.Abstract:
AI systems are increasingly applied to complex tasks that involve interaction with humans. During training, such systems are potentially dangerous, as they haven't yet learned to avoid actions that could cause serious harm. How can an AI system explore and learn without making a single mistake that harms humans or otherwise causes serious damage? For model-free reinforcement learning, having a human "in the loop" and ready to intervene is currently the only way to prevent all catastrophes. We formalize human intervention for RL and show how to reduce the human labor required by training a supervised learner to imitate the human's intervention decisions. We evaluate this scheme on Atari games, with a Deep RL agent being overseen by a human for four hours. When the class of catastrophes is simple, we are able to prevent all catastrophes without affecting the agent's learning (whereas an RL baseline fails due to catastrophic forgetting). However, this scheme is less successful when catastrophes are more complex: it reduces but does not eliminate catastrophes and the supervised learner fails on adversarial examples found by the agent. Extrapolating to more challenging environments, we show that our implementation would not scale (due to the infeasible amount of human labor required). We outline extensions of the scheme that are necessary if we are to train model-free agents without a single catastrophe.read more
Citations
More filters
Posted Content
Go-Explore: a New Approach for Hard-Exploration Problems
TL;DR: A new algorithm called Go-Explore, which exploits the following principles to remember previously visited states, solve simulated environments through any available means, and robustify via imitation learning, which results in a dramatic performance improvement on hard-exploration problems.
Posted Content
AI Safety Gridworlds
Jan Leike,Miljan Martic,Victoria Krakovna,Pedro A. Ortega,Tom Everitt,Andrew Lefrancq,Laurent Orseau,Shane Legg +7 more
TL;DR: A suite of reinforcement learning environments illustrating various safety properties of intelligent agents, including safe interruptibility, avoiding side effects, absent supervisor, reward gaming, safe exploration, as well as robustness to self-modification, distributional shift, and adversaries are presented.
Posted Content
Residual Reinforcement Learning for Robot Control.
Tobias Johannink,Shikhar Bahl,Ashvin Nair,Jianlan Luo,Avinash Kumar,Matthias Loskyll,Juan Aparicio Ojea,Eugen Solowjow,Sergey Levine +8 more
TL;DR: This paper studies how to solve difficult control problems in the real world by decomposing them into a part that is solved efficiently by conventional feedback control methods, and the residual which is solved with RL.
Posted Content
Scalable agent alignment via reward modeling: a research direction.
TL;DR: This work outlines a high-level research direction to solve the agent alignment problem centered around reward modeling: learning a reward function from interaction with the user and optimizing the learned reward function with reinforcement learning.
References
More filters
Journal ArticleDOI
Human-level control through deep reinforcement learning
Volodymyr Mnih,Koray Kavukcuoglu,David Silver,Andrei Rusu,Joel Veness,Marc G. Bellemare,Alex Graves,Martin Riedmiller,Andreas K. Fidjeland,Georg Ostrovski,Stig Petersen,Charles Beattie,Amir Sadik,Ioannis Antonoglou,Helen King,Dharshan Kumaran,Daan Wierstra,Shane Legg,Demis Hassabis +18 more
TL;DR: This work bridges the divide between high-dimensional sensory inputs and actions, resulting in the first artificial agent that is capable of learning to excel at a diverse array of challenging tasks.
Proceedings Article
Explaining and Harnessing Adversarial Examples
TL;DR: It is argued that the primary cause of neural networks' vulnerability to adversarial perturbation is their linear nature, supported by new quantitative results while giving the first explanation of the most intriguing fact about them: their generalization across architectures and training sets.
Proceedings Article
Asynchronous methods for deep reinforcement learning
Volodymyr Mnih,Adrià Puigdomènech Badia,Mehdi Mirza,Alex Graves,Tim Harley,Timothy P. Lillicrap,David Silver,Koray Kavukcuoglu +7 more
TL;DR: A conceptually simple and lightweight framework for deep reinforcement learning that uses asynchronous gradient descent for optimization of deep neural network controllers and shows that asynchronous actor-critic succeeds on a wide variety of continuous motor control problems as well as on a new task of navigating random 3D mazes using a visual input.
Deep reinforcement learning with double Q-learning
TL;DR: In this article, the authors show that the DQN algorithm suffers from substantial overestimation in some games in the Atari 2600 domain, and they propose a specific adaptation to the algorithm and show that this algorithm not only reduces the observed overestimations, but also leads to much better performance on several games.
Proceedings Article
Deep reinforcement learning with double Q-Learning
TL;DR: In this paper, the authors show that the DQN algorithm suffers from substantial overestimation in some games in the Atari 2600 domain, and they propose a specific adaptation to the algorithm and show that this algorithm not only reduces the observed overestimations, but also leads to much better performance on several games.