Analyzing Visual Representations in Embodied Navigation Tasks

Abstract

Recent advances in deep reinforcement learning require a large amount of training data and generally result in representations that are often over specialized to the target task. In this work, we present a methodology to study the underlying potential causes for this specialization. We use the recently proposed projection weighted Canonical Correlation Analysis (PWCCA) to measure the similarity of visual representations learned in the same environment by performing different tasks. We then leverage our proposed methodology to examine the task dependence of visual representations learned on related but distinct embodied navigation tasks. Surprisingly, we find that slight differences in task have no measurable effect on the visual representation for both SqueezeNet and ResNet architectures. We then empirically demonstrate that visual representations learned on one task can be effectively transferred to a different task.

References in corpus (9)

• SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
• High-Dimensional Continuous Control Using Generalized Advantage Estimation
• On Evaluation of Embodied Navigation Agents
• The Replica Dataset: A Digital Replica of Indoor Spaces
• Exploration by Random Network Distillation
• On the importance of single directions for generalization
• A Study on Overfitting in Deep Reinforcement Learning
• SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability
• Convergent Learning: Do different neural networks learn the same representations?