Randomized Value Functions via Multiplicative Normalizing Flows

Abstract

Randomized value functions offer a promising approach towards the challenge of efficient exploration in complex environments with high dimensional state and action spaces. Unlike traditional point estimate methods, randomized value functions maintain a posterior distribution over action-space values. This prevents the agent's behavior policy from prematurely exploiting early estimates and falling into local optima. In this work, we leverage recent advances in variational Bayesian neural networks and combine these with traditional Deep Q-Networks (DQN) and Deep Deterministic Policy Gradient (DDPG) to achieve randomized value functions for high-dimensional domains. In particular, we augment DQN and DDPG with multiplicative normalizing flows in order to track a rich approximate posterior distribution over the parameters of the value function. This allows the agent to perform approximate Thompson sampling in a computationally efficient manner via stochastic gradient methods. We demonstrate the benefits of our approach through an empirical comparison in high dimensional environments.

References in corpus (14)

• Continuous control with deep reinforcement learning
• Variational Inference with Normalizing Flows
• Weight Uncertainty in Neural Networks
• Density estimation using Real NVP
• Noisy Networks for Exploration
• VIME: Variational Information Maximizing Exploration
• Variational Dropout and the Local Reparameterization Trick
• Parameter Space Noise for Exploration
• Markov Chain Monte Carlo and Variational Inference: Bridging the Gap
• Incentivizing Exploration In Reinforcement Learning With Deep Predictive Models
• (More) Efficient Reinforcement Learning via Posterior Sampling
• Multiplicative Normalizing Flows for Variational Bayesian Neural Networks
• Deep Exploration via Randomized Value Functions
• Hierarchical Imitation and Reinforcement Learning

Cited by in corpus (9)

• Off-Policy Deep Reinforcement Learning without Exploration
• Benchmarking Bonus-Based Exploration Methods on the Arcade Learning Environment
• Worst-Case Regret Bounds for Exploration via Randomized Value Functions
• Attraction-Repulsion Actor-Critic for Continuous Control Reinforcement Learning
• Stochastic Neural Network with Kronecker Flow
• A Survey of Exploration Methods in Reinforcement Learning
• Randomised Bayesian Least-Squares Policy Iteration
• Parameterized Indexed Value Function for Efficient Exploration in Reinforcement Learning
• Out-of-Distribution Detection of Melanoma using Normalizing Flows