首页>
外国专利>
Self organising neural network formation and evaluation - using e.g. pattern of input and output electrodes, monitoring material changes due to discharge and using in learning process
Self organising neural network formation and evaluation - using e.g. pattern of input and output electrodes, monitoring material changes due to discharge and using in learning process
The method involves placing material under stress by physical or chemical means. Inhomogeneities produced in the material which can be evaluated. The artificial neural network is pref. represented by an electrical discharge through current conductive paths in an insulator (2). It learning process identifies discharges between input electrodes (3) and output electrode (4). The material in the paths changes due to the discharge effect and the output indicates the change which is monitored. A coupled circuit the changes to modify the input in a learning process. ADVANTAGE - Demonstrates neural network learning process.
展开▼