Perceptron Partition Model to Minimize Input Matrix

Abstract

Abstract Implementation of Neuron Network model using Perceptron has not given optimal result in real time learning. The large number of inputs expressed in matrix form makes the process slower in pattern recognition. So, it takes characteristic to represent all the input matrices by using the partition method. By partitioning each input and with the best weight and using the activation function will produce an output value. And learning is valid in recognizing patterns only 1 iteration only. Further validation is done on the water mill control module with dissolved oxygen input, water pH, salinity and water temperature. With Perceptron Partition learning algorithm more real-time than perceptron model. Testing on the waterwheel input whether rotating or stopping by Matrix Laboratory software simulation.