Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Same State, Different Task: Continual Reinforcement Learning without Interference

About

Continual Learning (CL) considers the problem of training an agent sequentially on a set of tasks while seeking to retain performance on all previous tasks. A key challenge in CL is catastrophic forgetting, which arises when performance on a previously mastered task is reduced when learning a new task. While a variety of methods exist to combat forgetting, in some cases tasks are fundamentally incompatible with each other and thus cannot be learnt by a single policy. This can occur, in reinforcement learning (RL) when an agent may be rewarded for achieving different goals from the same observation. In this paper we formalize this "interference" as distinct from the problem of forgetting. We show that existing CL methods based on single neural network predictors with shared replay buffers fail in the presence of interference. Instead, we propose a simple method, OWL, to address this challenge. OWL learns a factorized policy, using shared feature extraction layers, but separate heads, each specializing on a new task. The separate heads in OWL are used to prevent interference. At test time, we formulate policy selection as a multi-armed bandit problem, and show it is possible to select the best policy for an unknown task using feedback from the environment. The use of bandit algorithms allows the OWL agent to constructively re-use different continually learnt policies at different times during an episode. We show in multiple RL environments that existing replay based CL methods fail, while OWL is able to achieve close to optimal performance when training sequentially.

Samuel Kessler, Jack Parker-Holder, Philip Ball, Stefan Zohren, Stephen J. Roberts• 2021

Related benchmarks

TaskDatasetResultRank
Multi-agent Continual CooperationMAMuJoCo Reward (Expert)
Forward Transfer34.38
14
Dynamics task streamMaMuJoCo Medium
Backward Transfer2.97
8
Stalker-zealot task streamSMAC (Expert)
Backward Transfer21.65
8
Stalker-zealot task streamSMAC Medium
Backward Transfer11.31
8
Dynamics task streamMAMuJoCo (Expert)
Backward Transfer-0.2
8
Marines task streamSMAC (Expert)
Backward Transfer30.21
8
Marines task streamSMAC Medium
Backward Transfer9.64
8
Multi-agent Continual CooperationOverall All environments
Backward Transfer7.28
8
Reward task streamMAMuJoCo (Expert)
Backward Transfer19.05
8
Reward task streamMaMuJoCo Medium
Backward Transfer10.67
8
Showing 10 of 30 rows

Other info

Follow for update