Learning in the presence of concept drift and hidden contexts Export

@article{Widmer:96, abstract = {On-line learning in domains where the target concept depends on some hidden context poses serious problems. A changing context can induce changes in the target concepts, producing what is known as concept drift. We describe a family of learning algorithms that flexibly react to concept drift and can take advantage of situations where contexts reappear. The general approach underlying all these algorithms consists of (1) keeping only a window of currently trusted examples and hypotheses; (2) storing concept descriptions and reusing them when a previous context re-appears; and (3) controlling both of these functions by a heuristic that constantly monitors the system's behavior. The paper reports on experiments that test the systems' perfomance under various conditions such as different levels of noise and different extent and rate of concept drift.}, author = {Widmer, Gerhard and Kubat, Miroslav}, citeulike-article-id = {1720104}, citeulike-linkout-0 = {http://dx.doi.org/10.1007/BF00116900}, day = {1}, doi = {10.1007/BF00116900}, journal = {Machine Learning}, keywords = {concept\_drift, context, study}, month = {April}, number = {1}, pages = {69--101}, posted-at = {2008-04-08 17:24:37}, priority = {5}, title = {Learning in the presence of concept drift and hidden contexts}, url = {http://dx.doi.org/10.1007/BF00116900}, volume = {23}, year = {1996} }

TY - JOUR ID - Widmer:96 L3 - citeulike-article-id:1720104 N2 - On-line learning in domains where the target concept depends on some hidden context poses serious problems. A changing context can induce changes in the target concepts, producing what is known as concept drift. We describe a family of learning algorithms that flexibly react to concept drift and can take advantage of situations where contexts reappear. The general approach underlying all these algorithms consists of (1) keeping only a window of currently trusted examples and hypotheses; (2) storing concept descriptions and reusing them when a previous context re-appears; and (3) controlling both of these functions by a heuristic that constantly monitors the system's behavior. The paper reports on experiments that test the systems' perfomance under various conditions such as different levels of noise and different extent and rate of concept drift. IS - 1 JF - Machine Learning EP - 101 TI - Learning in the presence of concept drift and hidden contexts VL - 23 SP - 69 KW - concept_drift KW - context KW - study AU - Widmer, Gerhard AU - Kubat, Miroslav PY - 1996/04/01/ UR - http://dx.doi.org/10.1007/BF00116900 ER -