Interval Basis Neural Networks as a New Classification Tool

A. Horzyk (Poland)


: Ontogenic neural networks, deterministic interval basis configuration, classification.


The paper introduces a new classificator based on ontogenic neural networks called Interval Basis Neural Networks (IBNNs). The IBNN configures the whole topology and computes all weights after a priori knowledge collected from training data. IBNN proceeds statistical analyses of the training data for each class and produces intervals that group together training samples of the same class. The intervals are computed separately for all input features. This IBNNs feature makes possible to compute all network parameters without training. Moreover the IBNN takes into account the distances between patterns of the same class and builds the well-approximating model especially on the borders between the classes. The IBNNs are insensitive for differences in quantity of samples representing individual classes. The IBNNs always correctly classify training data and very good generalize other data.

Important Links:

Go Back