Scalable Bayesian Inference in the Era of Deep Learning: From Gaussian Processes to Deep Neural Networks

arXiv:2404.19157v1 Announce Type: cross
Abstract: Large neural networks trained on large datasets have become the dominant paradigm in machine learning. These systems rely on maximum likelihood point estimates of their parameters, precluding them from expressing model uncertainty. This may result in overconfident predictions and it prevents the use of deep learning models for sequential decision making. This thesis develops scalable methods to equip neural networks with model uncertainty. In particular, we leverage the linearised Laplace approximation to equip pre-trained neural …

abstract arxiv bayesian bayesian inference become cs.lg datasets deep learning gaussian processes inference large datasets likelihood machine machine learning maximum networks neural networks paradigm parameters predictions processes scalable stat.ml systems them type uncertainty