Arrangement for knowledge-based classification of datasets

摘要

An arrangement for parallel and therefore time-saving processing of data of any kind. In this arrangement, individual features of the data-capturing sensors (10) to (19) are present, and communicate by means of a knowledge base consisting of a matrix Mik with those elements (20) to (27) of a global classification level which contain the corresponding features. From the number and/or intensity of the signals which are output by the sensors, relative cumulative intensities or pulse counts are formed in registers (30) to (37) with the matrix Mik, and the register contents are compared to each other in a comparator (40). A signal output device (50) displays the address of the register which has reached the highest value. The arrangement is capable of learning, because the Miks are set or activated according to the recognition event which has previously been carried out. The arrangement is suitable, for example, for image recognition for diagnostic systems, and for observing sizes which change with time. IMAGE