SYSTEM AND METHOD FOR ON-LINE ADAPTIVE PREDICTION USING DYNAMIC MANAGEMENT OF MULTIPLE SUB-MODELS

    公开(公告)号:CA2344769A1

    公开(公告)日:2001-12-09

    申请号:CA2344769

    申请日:2001-04-20

    Applicant: IBM

    Abstract: Predictive models are widely used for tasks in many domains. The present invention addresses the problem of prediction of non-stationary processes by dynamical ly managing multiple models. The system comprises a model assessor, a model adapter, a plurality of sub-models, a plurality of model combiner functions, training data that is used to estimat e model parameters, and test data that is used to test for change points. Two processes are describe d, one for handling data updates and another that addresses prediction requests.

    On-line or real-time adaptive prediction

    公开(公告)号:GB2368425A

    公开(公告)日:2002-05-01

    申请号:GB0113938

    申请日:2001-06-08

    Applicant: IBM

    Abstract: A system for prediction of non-stationary processes by dynamically managing multiple models, comprises a model assessor, a model adapter, a plurality of sub-models, a plurality of model combiner functions, training data that is used to estimate model parameters, and test data that is used to test for change points. The model adapter selects sub-models to be combined to form a prediction on the basis of whether each sub-model has recently tested positive for a change point.

Patent Agency Ranking