Proceedings of the
The Nineteenth International Conference on Computational Intelligence and Security (CIS 2023)
December 1 – 4, 2023, Haikou, China
A Stochastic Variational Inference Based Bayesian Neural Network Algorithm
School of Mathematics and Statistics,Xidian University, China.
ABSTRACT
Bayesian Neural Network is powerful tools that provide uncertainty quantification and model robustness. However, traditional Bayesian Neural Network face challenges of high computational complexity and poor scalability during the inference process. To solve these problems, a new Bayesian Neural Network method based on variational inference is proposed. Our method utilizes stochastic variational inference to approximate the posterior distribution of the Bayesian Neural Network. Specifically, we employ a parameterized distribution to approximate the posterior distribution and optimize this approximation by maximizing the variational lower bound. Through this approach, We are able to effectively infer the weights of neural networks, quantify model uncertainty, accelerate the inference process, and generalize Bayesian Neural Network to large-scale datasets.
Keywords: Bayesian neural network, Variational inference, Stochastic optimization, Kullback-Leibler divergence.
Download PDF
School of Mathematics and Statistics,Xidian University, China.
ABSTRACT
Bayesian Neural Network is powerful tools that provide uncertainty quantification and model robustness. However, traditional Bayesian Neural Network face challenges of high computational complexity and poor scalability during the inference process. To solve these problems, a new Bayesian Neural Network method based on variational inference is proposed. Our method utilizes stochastic variational inference to approximate the posterior distribution of the Bayesian Neural Network. Specifically, we employ a parameterized distribution to approximate the posterior distribution and optimize this approximation by maximizing the variational lower bound. Through this approach, We are able to effectively infer the weights of neural networks, quantify model uncertainty, accelerate the inference process, and generalize Bayesian Neural Network to large-scale datasets.
Keywords: Bayesian neural network, Variational inference, Stochastic optimization, Kullback-Leibler divergence.
Download PDF