In the present paper nonholonomic systems with drift terms are studied. The discussion is focused on a class of Lagrangian systems with a cyclic coordinate. We present an approach...
J.-M. Godhavn, A. Balluchi, L. S. Crawford, S. S. ...
We review the application of statistical mechanics methods to the study of online learning of a drifting concept in the limit of large systems. The model where a feed-forward netwo...
Mining evolving data streams for concept drifts has gained importance in applications like customer behavior analysis, network intrusion detection, credit card fraud detection. Se...
Most of supervised learning algorithms assume the stability of the target concept over time. Nevertheless in many real-user modeling systems, where the data is collected over an ex...
To cope with concept drift, we paired a stable online learner with a reactive one. A stable learner predicts based on all of its experience, whereas a reactive learner predicts ba...