The Utility Of Sparse Representations For Control In Reinforcement Learning

Abstract

We investigate sparse representations for control in reinforcement learning. While these representations are widely used in computer vision, their prevalence in reinforcement learning is limited to sparse coding where extracting representations for new data can be computationally intensive. Here, we begin by demonstrating that learning a control policy incrementally with a representation from a standard neural network fails in classic control domains, whereas learning with a representation obtained from a neural network that has sparsity properties enforced is effective. We provide evidence that the reason for this is that the sparse representation provides locality, and so avoids catastrophic interference, and particularly keeps consistent, stable values for bootstrapping. We then discuss how to learn such sparse representations. We explore the idea of Distributional Regularizers, where the activation of hidden nodes is encouraged to match a particular distribution that results in spa

Stats

• citations24
• S2 citations—
• github stars0
• HF likes0
• heat score10.48
• arxiv keyliu2018the

Related papers

• Learning Sparse Representations In Reinforcement Learning With Sparse Coding (2017)8.82
• Learning Sparse Representations Incrementally In Deep Reinforcement Learning (2019)0.00
• Optimal Control With Natural Images: Efficient Reinforcement Learning Using Overcomplete Sparse Codes (2024)0.00
• Reinforcement Learning With Sparse-executing Actions Via Sparsity Regularization (2021)0.00
• Dynamic Sparsity: Challenging Common Sparsity Assumptions For Learning World Models In Robotic Reinforcement Learning Benchmarks (2025)0.00
• Dynamic Sparse Training For Deep Reinforcement Learning (2021)0.00
• Improving Sample Efficiency In Model-free Reinforcement Learning From Images (2019)16.99
• Data-efficient Reinforcement Learning With Self-predictive Representations (2020)0.00