Low-rank extended Kalman filtering for online learning of neural networks from streaming data
This work addresses the need for efficient online learning methods for neural networks in non-stationary environments, representing an incremental improvement over existing approaches.
The paper tackles the problem of online learning from streaming data by proposing a low-rank extended Kalman filter algorithm for efficient Bayesian inference, resulting in faster learning and adaptation with improved sample efficiency and reward accumulation in contextual bandits.
We propose an efficient online approximate Bayesian inference algorithm for estimating the parameters of a nonlinear function from a potentially non-stationary data stream. The method is based on the extended Kalman filter (EKF), but uses a novel low-rank plus diagonal decomposition of the posterior precision matrix, which gives a cost per step which is linear in the number of model parameters. In contrast to methods based on stochastic variational inference, our method is fully deterministic, and does not require step-size tuning. We show experimentally that this results in much faster (more sample efficient) learning, which results in more rapid adaptation to changing distributions, and faster accumulation of reward when used as part of a contextual bandit algorithm.