Pre-training with Synthetic Data Helps Offline Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Zecheng, Wang, Che, Dong, Zixuan, Ross, Keith
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914812305342464
author Wang, Zecheng
Wang, Che
Dong, Zixuan
Ross, Keith
author_facet Wang, Zecheng
Wang, Che
Dong, Zixuan
Ross, Keith
contents Recently, it has been shown that for offline deep reinforcement learning (DRL), pre-training Decision Transformer with a large language corpus can improve downstream performance (Reid et al., 2022). A natural question to ask is whether this performance gain can only be achieved with language pre-training, or can be achieved with simpler pre-training schemes which do not involve language. In this paper, we first show that language is not essential for improved performance, and indeed pre-training with synthetic IID data for a small number of updates can match the performance gains from pre-training with a large language corpus; moreover, pre-training with data generated by a one-step Markov chain can further improve the performance. Inspired by these experimental results, we then consider pre-training Conservative Q-Learning (CQL), a popular offline DRL algorithm, which is Q-learning-based and typically employs a Multi-Layer Perceptron (MLP) backbone. Surprisingly, pre-training with simple synthetic data for a small number of updates can also improve CQL, providing consistent performance improvement on D4RL Gym locomotion datasets. The results of this paper not only illustrate the importance of pre-training for offline DRL but also show that the pre-training data can be synthetic and generated with remarkably simple mechanisms.
format Preprint
id arxiv_https___arxiv_org_abs_2310_00771
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Pre-training with Synthetic Data Helps Offline Reinforcement Learning
Wang, Zecheng
Wang, Che
Dong, Zixuan
Ross, Keith
Artificial Intelligence
Machine Learning
Recently, it has been shown that for offline deep reinforcement learning (DRL), pre-training Decision Transformer with a large language corpus can improve downstream performance (Reid et al., 2022). A natural question to ask is whether this performance gain can only be achieved with language pre-training, or can be achieved with simpler pre-training schemes which do not involve language. In this paper, we first show that language is not essential for improved performance, and indeed pre-training with synthetic IID data for a small number of updates can match the performance gains from pre-training with a large language corpus; moreover, pre-training with data generated by a one-step Markov chain can further improve the performance. Inspired by these experimental results, we then consider pre-training Conservative Q-Learning (CQL), a popular offline DRL algorithm, which is Q-learning-based and typically employs a Multi-Layer Perceptron (MLP) backbone. Surprisingly, pre-training with simple synthetic data for a small number of updates can also improve CQL, providing consistent performance improvement on D4RL Gym locomotion datasets. The results of this paper not only illustrate the importance of pre-training for offline DRL but also show that the pre-training data can be synthetic and generated with remarkably simple mechanisms.
title Pre-training with Synthetic Data Helps Offline Reinforcement Learning
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2310.00771