Learning in Time Ordered Domains with Hidden Changes in Context

Concept drift due to hidden changes in context complicates learning in many real world domains including financial prediction, medical diagnosis, and communication network performance. Machine learning systems addressing this problem generally use an incremental learning, on-line paradigm. An off-line, meta-learning approach to the identification of hidden context is presented. This approach uses an existing batch learner and the process of contextual clustering to identify stable hidden contexts, and the associated, context specific, locally stable concepts. The approach is broadly applicable to a range of domains and learning methods. We describe several evaluation domains and report current progress on these domains.