Disentangled (Un)Controllable Features

Abstract

In the context of MDPs with high-dimensional states, downstream tasks are predominantly applied on a compressed, low-dimensional representation of the original input space. A variety of learning objectives have therefore been used to attain useful representations. However, these representations usually lack interpretability of the different features. We present a novel approach that is able to disentangle latent features into a controllable and an uncontrollable partition. We illustrate that the resulting partitioned representations are easily interpretable on three types of environments and show that, in a distribution of procedurally generated maze environments, it is feasible to interpretably employ a planning algorithm in the isolated controllable latent partition.

14 pages (8 main paper pages), 15 figures

References in corpus (9)

• Generative Adversarial Networks
• DeepMind Control Suite
• Contrastive Learning of Structured World Models
• Independently Controllable Factors
• Predictive Information Accelerates Learning in RL
• A Survey on Interpretable Reinforcement Learning
• Weakly Supervised Representation Learning with Sparse Perturbations
• Denoised MDPs: Learning World Models Better Than the World Itself
• Feature-Based Interpretable Reinforcement Learning based on State-Transition Models

Cited by in corpus (1)

• Structure in Deep Reinforcement Learning: A Survey and Open Problems