Co-Adaptation of Algorithmic and Implementational Innovations in Inference-based Deep Reinforcement Learning

TL;DR

This paper disentangles algorithmic innovations from implementation details in inference-based deep RL algorithms, revealing which aspects are co-adapted and transferable, thereby clarifying sources of performance improvements.

Contribution

It provides a unified derivation framework for off-policy inference-based RL algorithms and systematically analyzes the impact of implementation choices on performance.

Findings

Implementation details significantly affect algorithm performance.

Certain implementation choices are highly co-adapted with specific algorithms.

Some implementation techniques transfer effectively across algorithms.

Abstract

Recently many algorithms were devised for reinforcement learning (RL) with function approximation. While they have clear algorithmic distinctions, they also have many implementation differences that are algorithm-independent and sometimes under-emphasized. Such mixing of algorithmic novelty and implementation craftsmanship makes rigorous analyses of the sources of performance improvements across algorithms difficult. In this work, we focus on a series of off-policy inference-based actor-critic algorithms -- MPO, AWR, and SAC -- to decouple their algorithmic innovations and implementation decisions. We present unified derivations through a single control-as-inference objective, where we can categorize each algorithm as based on either Expectation-Maximization (EM) or direct Kullback-Leibler (KL) divergence minimization and treat the rest of specifications as implementation details. We performed extensive ablation studies, and identified substantial performance drops whenever implementation details are mismatched for algorithmic choices. These results show which implementation or code details are co-adapted and co-evolved with algorithms, and which are transferable across algorithms: as examples, we identified that tanh Gaussian policy and network sizes are highly adapted to algorithmic types, while layer normalization and ELU are critical for MPO's performances but also transfer to noticeable gains in SAC. We hope our work can inspire future work to further demystify sources of performance improvements across multiple algorithms and allow researchers to build on one another's both algorithmic and implementational innovations.