Detecting and Adapting to Irregular Distribution Shifts in Bayesian Online Learning
arXiv:2012.08101
Abstract
We consider the problem of online learning in the presence of distribution shifts that occur at an unknown rate and of unknown intensity. We derive a new Bayesian online inference approach to simultaneously infer these distribution shifts and adapt the model to the detected changes by integrating ideas from change point detection, switching dynamical systems, and Bayesian online learning. Using a binary 'change variable,' we construct an informative prior such that--if a change is detected--the model partially erases the information of past model updates by tempering to facilitate adaptation to the new data distribution. Furthermore, the approach uses beam search to track multiple change-point hypotheses and selects the most probable one in hindsight. Our proposed method is model-agnostic, applicable in both supervised and unsupervised learning settings, suitable for an environment of concept drifts or covariate drifts, and yields improvements over state-of-the-art Bayesian online learning approaches.
arXiv commentPublished version, Neural Information Processing Systems 2021
References in corpus (17)
- Distributed Representations of Words and Phrases and their Compositionality
- Auto-Encoding Variational Bayes
- Overcoming catastrophic forgetting in neural networks
- Confounding variables can degrade generalization performance of radiological deep learning models
- Continual Learning Through Synaptic Intelligence
- Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift
- Bayesian Online Changepoint Detection
- Dynamic Word Embeddings
- Changepoint detection for high-dimensional time series with missing data
- A Neural Dirichlet Process Mixture Model for Task-Free Continual Learning
- Functional Regularisation for Continual Learning with Gaussian Processes
- Continual Unsupervised Representation Learning
- Structured Stochastic Variational Inference
- A Unifying Bayesian View of Continual Learning
- Switching Linear Dynamics for Variational Bayes Filtering
- Spatio-temporal Bayesian On-line Changepoint Detection with Model Selection
- A trust-region method for stochastic variational inference with applications to streaming data