Abstract: An artificial vision system traning method starts by forming (S2) a feature matrix including feature sample vectors and a corresponding response sample vector including response sample scalars. The method then uses an iterative procedure to determine a linkage vector linking the response sample vector to the feature matrix. This iterative procedure includes the steps: determining (S4) a response sample vector error estimate in the response sample vector domain; transforming (S6) the response sample vector error estimate into a corresponding linkage vector error estimate in the linkage vector domain; determining (S7) a linkage vector estimate in the linkage vector domain by using the linkage vector error estimate; transforming (S8) the linkage vector estimate into a corresponding response sample vector estimate in the response sample vector domain. These steps are repeated until (S5) the response sample vector error estimate is sufficiently small.

Abstract: An invention forcing an aggregate risk model to be consistent with standalone models is provided. A revising transformation parameterized over and an objective function minimized over, the orthogonal group are provided, least changing cross blocks of covariance matrices, preserving information in original cross block correlations, consistent with a prescribed revised sub-block.

Abstract: The present invention provides techniques for transmitting at least one signal through an element of a classification system. One or more input signals are received at the element. One or more functional components are extracted from the one or more input signals, and one or more membership components are extracted from the one or more input signals. An output signal is generated from the element comprising a functional component and a membership component that correspond to one or more functional components and membership components from one or more input signals.

Abstract: A predictive model method (and structure) includes receiving an input data into an initial model to develop an initial model output and receiving both of the input data and the initial model output as input data into a first transform/regression stage.

Abstract: Selection of certain attributes as output and input attributes is provided so a decision tree may be created more efficiently. For each possible output attribute an interestingness score is calculated. This interestingness score is based on entropy of the output attribute and a desirable entropy constant. The attributes with the highest interestingness score are used as output attributes in the creation of the decision tree. Score gains for the input attribute over the output attributes are calculated using a conventional scoring algorithm. The sum of the score gains over all output attributes for each input attribute is calculated. The attributes with the highest score gain sums are used as input attributes in the creation of the decision tree.

Abstract: According to a first aspect of the present invention there is provided a method of modelling a network comprising operating the network as a neural network and executing a neural network modelling algorithm on the network, whereby the network models its own response to a requested action.