Instance based Generalization in Reinforcement Learning

by   Martin Bertran, et al.

Agents trained via deep reinforcement learning (RL) routinely fail to generalize to unseen environments, even when these share the same underlying dynamics as the training levels. Understanding the generalization properties of RL is one of the challenges of modern machine learning. Towards this goal, we analyze policy learning in the context of Partially Observable Markov Decision Processes (POMDPs) and formalize the dynamics of training levels as instances. We prove that, independently of the exploration strategy, reusing instances introduces significant changes on the effective Markov dynamics the agent observes during training. Maximizing expected rewards impacts the learned belief state of the agent by inducing undesired instance specific speedrunning policies instead of generalizeable ones, which are suboptimal on the training set. We provide generalization bounds to the value gap in train and test environments based on the number of training instances, and use insights based on these to improve performance on unseen levels. We propose training a shared belief representation over an ensemble of specialized policies, from which we compute a consensus policy that is used for data collection, disallowing instance specific exploitation. We experimentally validate our theory, observations, and the proposed computational solution over the CoinRun benchmark.



page 1

page 2

page 3

page 4


Robust Deep Reinforcement Learning for Quadcopter Control

Deep reinforcement learning (RL) has made it possible to solve complex r...

Learning Domain Invariant Representations in Goal-conditioned Block MDPs

Deep Reinforcement Learning (RL) is successful in solving many complex M...

Active Reinforcement Learning over MDPs

The past decade has seen the rapid development of Reinforcement Learning...

Exploring the Memorization-Generalization Continuum in Deep Learning

Human learners appreciate that some facts demand memorization whereas ot...

Self-Paced Context Evaluation for Contextual Reinforcement Learning

Reinforcement learning (RL) has made a lot of advances for solving a sin...

Solving Sokoban with backward reinforcement learning

In some puzzles, the strategy we need to use near the goal can be quite ...

Provably Efficient Causal Model-Based Reinforcement Learning for Systematic Generalization

In the sequential decision making setting, an agent aims to achieve syst...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.