ADAPTIVE PROCESSING PARAMETER ADJUSTMENT BY FEEDBACK RECOGNITION METHOD WITH INVERSE RECALL NEURAL-NETWORK MODEL

A feedback pattern recognition method using an inverse recall neural n etwork model is proposed. The feedback method can adjust processing pa rameter values adaptively to individual patterns so as to produce reli able recognition results. In order to apply an adaptive control techni que to such pattern recognition processings, the evaluation value for recognition uncertainty is determined to be a function with regard to an input pattern and processing parameters. In its feedback phase, the input pattern is fixed and processing parameters are adjusted to decr ease the recognition uncertainty. The proposed neural network model im plements two functions in this feedback recognition method. One is a d iscrimination as a kind of multi-layer feedforward model. The other is to generate an input modification so as to decrease the recognition u ncertainty. The modification values indicate parts which are important for more certain recognition but are missed in the original input to the network. The proposed feedback method can adjust processing parame ter values in order to detect the important parts shown by the inverse recall network model. As explained in this paper, feature extraction parameter values are adaptively adjusted by this feedback method. Afte r the inverse recall model and the feedback function are implemented, features are extracted again by using the modified feature extraction parameter values. The feature is classified by the feedforward functio n of the network model. The feedforward and feedback processings are r epeated until a certain recognition result is obtained. This method wa s examined for handwritten alpha-numerics with rotation distortion. Th e feedback method was found to decrease the rejection ratio at the sam e substitution error ratio with high efficiency.