Perceptron.java

import java.text.DecimalFormat;
import java.util.Random;

public class Perceptron {
	public static void main (String[] args) {
		double[][] inputs = {
			{1, 1},
			{1, 0},
			{0, 1},
			{0, 0}
		};

		double[] outputs = {
			1,
			0,
			0,
			0
		};

		Perceptron p = new Perceptron(2);

		DecimalFormat df = new DecimalFormat("0.000000");
		System.out.println("untrained");
		for(double[] input : inputs){
			double guess = p.guess(input);
			System.out.println(df.format(guess));
		}

		Random random = new Random(System.nanoTime());

		for (int i = 0; i < 10000; i++) {
			int index = random.nextInt(4);
			p.train(inputs[index], outputs[index]);
		}

		System.out.println("trained");
		for(double[] input : inputs){
			double guess = p.guess(input);
			System.out.println(df.format(guess));
		}
	}

	double[] weights;
	double learningRate = 0.1;

	public Perceptron(int inputCount){
		this.weights = new double[inputCount+1];

		for (int i = 0; i < weights.length; i++) {
			weights[i] = .5f;
		}
	}

	public double guess(double[] input){
		double sum = 0;
		for (int i = 0; i < input.length; i++) {
			sum += input[i] * weights[i];
		}
		// BIAS
		sum += weights[weights.length-1];

		return this.activation(sum);
	}

	public void train(double[] input, double label){
		double guess = this.guess(input);
		double error = label - guess;

		for (int i = 0; i < input.length; i++) {
			this.weights[i] += error * input[i] * this.learningRate;
		}
		int lastIndex = weights.length - 1;
		this.weights[lastIndex] += error * this.learningRate;
	}

	public double activation(double n){
		return 1d/(1d + Math.exp(-n));
	}
}

kozak