Dynamic logistic regression
- 22 January 2003
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 3 (10987576) , 1562-1567
- https://doi.org/10.1109/ijcnn.1999.832603
Abstract
We propose an online learning algorithm for training a logistic regression model on nonstationary classification problems. The nonstationarity is captured by modelling the weights in a logistic regression classifier as evolving according to a first order Markov process. The weights are updated using the extended Kalman filter formalism and nonstationarities are tracked by inferring a time-varying state noise variance parameter. We describe an algorithm for doing this based on maximising the evidence of updated predictions. The algorithm is illustrated on a number of synthetic problems.Keywords
This publication has 7 references indexed in Scilit:
- A Unifying Review of Linear Gaussian ModelsNeural Computation, 1999
- Neural Networks for Pattern RecognitionPublished by Oxford University Press (OUP) ,1995
- A recursive multiple model approach to noise identificationIEEE Transactions on Aerospace and Electronic Systems, 1994
- The Evidence Framework Applied to Classification NetworksNeural Computation, 1992
- Sequential updating of conditional probabilities on directed graphical structuresNetworks, 1990
- Dynamic Generalized Linear Models and Bayesian Forecasting: RejoinderJournal of the American Statistical Association, 1985
- Adaptive filteringAutomatica, 1969