Avoiding Side Effects in Complex Environments
arXiv:2006.06547
Abstract
Reward function specification can be difficult. Rewarding the agent for making a widget may be easy, but penalizing the multitude of possible negative side effects is hard. In toy environments, Attainable Utility Preservation (AUP) avoided side effects by penalizing shifts in the ability to achieve randomly generated goals. We scale this approach to large, randomly generated environments based on Conway's Game of Life. By preserving optimal value for a single randomly generated reward function, AUP incurs modest overhead while leading the agent to complete the specified task and avoid many side effects. Videos and code are available at https://avoiding-side-effects.github.io/.
arXiv commentAccepted as spotlight paper at NeurIPS 2020. 10 pages main paper; 19 pages with appendices
References in corpus (9)
- Deep reinforcement learning from human preferences
- Safe Model-based Reinforcement Learning with Stability Guarantees
- Benchmarking Batch Deep Reinforcement Learning Algorithms
- AI Safety Gridworlds
- A Lyapunov-based Approach to Safe Reinforcement Learning
- Inverse Reward Design
- The continuous Bernoulli: fixing a pervasive error in variational autoencoders
- SafeLife 1.0: Exploring Side Effects in Complex Environments
- Towards Empathic Deep Q-Learning