🤖 AI Summary
This study addresses the uncontrollable error and computational inefficiency arising from fixed sampling budgets in Monte Carlo inference for Bayesian neural networks (BNNs). To this end, it introduces confidence sequence theory into BNN inference for the first time, proposing an adaptive sampling termination algorithm. This method dynamically determines the required number of samples based on specific decision-making objectives, such as classification or distribution approximation, enabling on-demand stopping with rigorous statistical guarantees. Experimental results demonstrate that the proposed mechanism intelligently allocates computational resources by automatically increasing the sample count for ambiguous inputs. Consequently, it significantly reduces overall inference latency while preserving decision reliability.
📝 Abstract
Bayesian neural network predictions are commonly approximated using a fixed number of Monte Carlo samples per input, without controlling the resulting error that comes from this finite sample. We propose the use of confidence sequences to dynamically determine how many samples are needed while maintaining statistical guarantees. We consider several ways in which predictive probabilities are used, including identifying the most likely class, approximating the full predictive distribution, and resolving probability-threshold decisions. Sampling stops once the corresponding decision can be made with the desired guarantee. Experiments show that the method allocates the computational budget efficiently, assigning more samples to ambiguous inputs than to easy inputs while preserving reliable decisions and reducing overall latency relative to a fixed Monte Carlo budget.