Privileged Information Dropout in Reinforcement Learning

2020/05/19 by Pierre-Alexandre Kamienny, Kamienny, Pierre-Alexandre, Kai Arulkumaran +7 · 2 citations
Computer Science · #Artificial Intelligence (cs.AI) #Data Stream Mining Techniques #Evolutionary Algorithms and Applications #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Reinforcement Learning in Robotics

paper · pdf · doi:10.48550/arxiv.2005.09220

openalex publication_date 2020/05/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Using privileged information during training can improve the sample efficiency and performance of machine learning systems. This paradigm has been applied to reinforcement learning (RL), primarily in the form of distillation or auxiliary tasks, and less commonly in the form of augmenting the inputs of agents. In this work, we investigate Privileged Information Dropout (\pid) for achieving the latter which can be applied equally to value-based and policy-based RL algorithms. Within a simple partially-observed environment, we demonstrate that \pid outperforms alternatives for leveraging privileged information, including distillation and auxiliary tasks, and can successfully utilise different types of privileged information. Finally, we analyse its effect on the learned representations.

Citations

• The information bottleneck method
• Distilling the Knowledge in a Neural Network
• A Survey on Transfer Learning
• Parameter Space Noise for Exploration
• Noisy Networks for Exploration
• Counterfactual Multi-Agent Policy Gradients

Cited by

• Driving Beyond Privilege: Distilling Dense-Reward Knowledge into Sparse-Reward Policies
• Unifying Agent Interaction and World Information for Multi-agent Coordination

Related

• To Distill or Decide? Understanding the Algorithmic Trade-off in Partially Observable Reinforcement Learning
• PIGDreamer: Privileged Information Guided World Models for Safe Partially Observable Reinforcement Learning
• Provable Partially Observable Reinforcement Learning with Privileged Information
• Bridging the Sim-to-Real Gap from the Information Bottleneck Perspective
• Robust Predictable Control