LGOct 19, 2021

Balancing Value Underestimation and Overestimation with Realistic Actor-Critic

arXiv:2110.09712v64 citations
Originality Highly original
AI Analysis

This addresses a key bottleneck in applying RL to real-world domains by improving sample efficiency, though it is incremental as it builds on existing off-policy methods.

The paper tackles the problem of poor sample efficiency in model-free deep reinforcement learning for continuous control by introducing Realistic Actor-Critic (RAC), which uses Universal Value Function Approximators and uncertainty punished Q-learning to learn a policy family balancing value underestimation and overestimation, achieving 10x sample efficiency and 25% performance improvement on the Humanoid environment compared to SAC.

Model-free deep reinforcement learning (RL) has been successfully applied to challenging continuous control domains. However, poor sample efficiency prevents these methods from being widely used in real-world domains. This paper introduces a novel model-free algorithm, Realistic Actor-Critic(RAC), which can be incorporated with any off-policy RL algorithms to improve sample efficiency. RAC employs Universal Value Function Approximators (UVFA) to simultaneously learn a policy family with the same neural network, each with different trade-offs between underestimation and overestimation. To learn such policies, we introduce uncertainty punished Q-learning, which uses uncertainty from the ensembling of multiple critics to build various confidence-bounds of Q-function. We evaluate RAC on the MuJoCo benchmark, achieving 10x sample efficiency and 25\% performance improvement on the most challenging Humanoid environment compared to SAC.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes