Pemanfaatan Algoritma Pembelajaran Pola Karakter Menggunakan Metode Hebb Rule

Abstract

The neuron model will experience difficulties when it encounters complex functions, this is because determining the weight w and threshold value ø must be processed analytically. The Heb Rule method finds a way to calculate the weight value w and bias value which can be processed easily without carrying out a training process first. The Hebb Rule algorithm is the oldest method with a work process using learning and supervision methods. The Hebb Rulu architecture is the same as the McCulloch-Pitts network architecture in that several input units are connected to output units, equipped with a bias value. The problem that often occurs in research is the difference between the Hebb Rule network value pattern and the specified target value due to the incorrect weight initialization process in the calculation process. The aim of the research is to obtain pattern results in accordance with the predetermined targets. Based on the input pattern which consists of pattern 1, pattern 2, pattern 3 and pattern 4, all patterns can be read by the system. The final pattern results are, P1=1, P2=2, P3=1, P4=1.