Temporal shift reinforcement learning
- 5 April 2022
- conference paper
- conference paper
- Published by Association for Computing Machinery (ACM) in Proceedings of the 2nd European Workshop on Machine Learning and Systems
Abstract
The function approximators employed by traditional image-based Deep Reinforcement Learning (DRL) algorithms usually lack a temporal learning component and instead focus on learning the spatial component. We propose a technique, Temporal Shift Reinforcement Learning (TSRL), wherein both temporal, as well as spatial components are jointly learned. Moreover, TSRL does not require additional parameters to perform temporal learning. We show that TSRL outperforms the commonly used frame stacking heuristic on all of the Atari environments we test on while beating the SOTA for all except one of them. This investigation has implications in the robotics as well as sequential decision-making domains. Our code is available at - https://github.com/Deepakgthomas/TSM_RLKeywords
This publication has 11 references indexed in Scilit:
- TSM: Temporal Shift Module for Efficient Video UnderstandingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2019
- Learning Deep Visuomotor Policies for Dexterous Hand ManipulationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2019
- ECO: Efficient Convolutional Network for Online Video UnderstandingPublished by Springer Science and Business Media LLC ,2018
- Temporal Relational Reasoning in VideosPublished by Springer Science and Business Media LLC ,2018
- Quo Vadis, Action Recognition? A New Model and the Kinetics DatasetPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2017
- Path integral guided policy searchPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2017
- Temporal Segment Networks: Towards Good Practices for Deep Action RecognitionPublished by Springer Science and Business Media LLC ,2016
- Learning Spatiotemporal Features with 3D Convolutional NetworksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2015
- Human-level control through deep reinforcement learningNature, 2015
- Large-Scale Video Classification with Convolutional Neural NetworksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2014