Kalman Filter for Online Classification of Non-Stationary Data

by   Michalis K. Titsias, et al.

In Online Continual Learning (OCL) a learning system receives a stream of data and sequentially performs prediction and training steps. Important challenges in OCL are concerned with automatic adaptation to the particular non-stationary structure of the data, and with quantification of predictive uncertainty. Motivated by these challenges we introduce a probabilistic Bayesian online learning model by using a (possibly pretrained) neural representation and a state space model over the linear predictor weights. Non-stationarity over the linear predictor weights is modelled using a parameter drift transition density, parametrized by a coefficient that quantifies forgetting. Inference in the model is implemented with efficient Kalman filter recursions which track the posterior distribution over the linear weights, while online SGD updates over the transition dynamics coefficient allows to adapt to the non-stationarity seen in data. While the framework is developed assuming a linear Gaussian model, we also extend it to deal with classification problems and for fine-tuning the deep learning representation. In a set of experiments in multi-class classification using data sets such as CIFAR-100 and CLOC we demonstrate the predictive ability of the model and its flexibility to capture non-stationarity.


page 1

page 2

page 3

page 4


Kalman Filter Modifier for Neural Networks in Non-stationary Environments

Learning in a non-stationary environment is an inevitable problem when a...

Online Learning of the Kalman Filter with Logarithmic Regret

In this paper, we consider the problem of predicting observations genera...

Low-rank extended Kalman filtering for online learning of neural networks from streaming data

We propose an efficient online approximate Bayesian inference algorithm ...

Continual Learning with Pretrained Backbones by Tuning in the Input Space

The intrinsic difficulty in adapting deep learning models to non-station...

Non-Linear Non-Stationary Heteroscedasticity Volatility for Tracking of Jump Processes

In this paper, we introduce a new jump process modeling which involves a...

Regularization of the ensemble Kalman filter using a non-stationary spatial convolution model

Applications of the ensemble Kalman filter to high-dimensional problems ...

A Variational Bayesian State-Space Approach to Online Passive-Aggressive Regression

Online Passive-Aggressive (PA) learning is a class of online margin-base...

Please sign up or login with your details

Forgot password? Click here to reset