Learning Structural Weight Uncertainty for Sequential Decision-Making
This work addresses the need for better uncertainty quantification in neural networks for sequential decision-making problems like contextual bandits and reinforcement learning, offering an incremental improvement by incorporating structural priors.
The paper tackled the problem of learning structural weight uncertainty in neural networks for sequential decision-making, proposing a method using matrix variate Gaussian priors within an SVGD framework, and demonstrated superiority over state-of-the-art methods in experiments on synthetic and real datasets.
Learning probability distributions on the weights of neural networks (NNs) has recently proven beneficial in many applications. Bayesian methods, such as Stein variational gradient descent (SVGD), offer an elegant framework to reason about NN model uncertainty. However, by assuming independent Gaussian priors for the individual NN weights (as often applied), SVGD does not impose prior knowledge that there is often structural information (dependence) among weights. We propose efficient posterior learning of structural weight uncertainty, within an SVGD framework, by employing matrix variate Gaussian priors on NN parameters. We further investigate the learned structural uncertainty in sequential decision-making problems, including contextual bandits and reinforcement learning. Experiments on several synthetic and real datasets indicate the superiority of our model, compared with state-of-the-art methods.