BayesAdapter: Being Bayesian, Inexpensively and Robustly, via Bayesian Fine-tuning

Zhijie Deng, Xiao Yang, Hao Zhang, Yinpeng Dong, Jun Zhu

Keywords: Bayesian neural networks, Bayesian fine-tuning, uncertainty estimation, OOD detection

Abstract: Despite their theoretical appealingness, Bayesian neural networks (BNNs) are falling far behind in terms of adoption in real-world applications compared with normal NNs, mainly due to their limited scalability in training, and low fidelity in their uncertainty estimates. In this work, we develop a new framework, named BayesAdapter, to address these issues and bring Bayesian deep learning to the masses. The core notion of BayesAdapter is to adapt pre-trained deterministic NNs to be BNNs via Bayesian fine-tuning. We implement Bayesian fine-tuning with a plug-and-play instantiation of stochastic variational inference, and propose exemplar reparameterization to reduce gradient variance and stabilize the fine-tuning. Together, they enable training BNNs as if one were training deterministic NNs with minimal added overheads. During Bayesian fine-tuning, we further propose an uncertainty regularization to supervise and calibrate the uncertainty quantification of learned BNNs at low cost. To empirically evaluate BayesAdapter, we conduct extensive experiments on a diverse set of challenging benchmarks, and observe satisfactory training efficiency, competitive predictive performance, and calibrated and faithful uncertainty estimates.

One-sentence Summary: We propose to obtain a reliable BNN by fine-tuning a pre-trained DNN under uncertainty regularization with minimal added overheads.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=4rpenXQyc