com.sankalp.characterreader.CharacterReader.java Source code

Introduction

Here is the source code for com.sankalp.characterreader.CharacterReader.java

Source

/*
 * To change this license header, choose License Headers in Project Properties.
 * To change this template file, choose Tools | Templates
 * and open the template in the editor.
 */

package com.sankalp.characterreader;

import com.sankalp.characterreader.neuralnetwork.TrainingData;
import com.sankalp.characterreader.neuralnetwork.NeuralNetwork;

import java.io.BufferedReader;
import java.io.File;
import java.io.IOException;
import java.io.InputStreamReader;
import java.util.ArrayList;
import java.util.List;

import org.apache.commons.io.FileUtils;

/**
 *
 * @author sankalpkulshrestha
 */
public class CharacterReader {

    public static void main(String[] args) throws IOException {

        double learningRate = 0.55;
        List<NeuralNetwork.Layer> hiddenLayerList = new ArrayList();
        hiddenLayerList.add(new NeuralNetwork.Layer(50));
        NeuralNetwork neuralNetwork = new NeuralNetwork(new NeuralNetwork.Layer(28 * 28),
                new NeuralNetwork.Layer(10), hiddenLayerList, learningRate);

        int trainingSampleCount = 60000;
        TrainingData trainingData = new TrainingData(new File("/home/sankalpkulshrestha/mnist/train-images/"),
                trainingSampleCount, new File("/home/sankalpkulshrestha/mnist/train-labels.csv"));

        int totalEpochs = 30;
        for (int index = 0; index < totalEpochs; index++) {
            System.out.println("---------- EPOCH " + index + " ----------");
            neuralNetwork.train(trainingData);

            TrainingData testData = new TrainingData(new File("/home/sankalpkulshrestha/mnist/test-images/"), 10000,
                    new File("/home/sankalpkulshrestha/mnist/test-labels.csv"));

            System.out.println("Training over");
            double accuracy = neuralNetwork.measureAccuracy(testData);
            System.out.println(accuracy);
            if (index < totalEpochs - 1) {
                trainingData.shuffle();
                trainingData.reset();
            }
        }
        System.out.println("Model trained. Enter image file names:");
        BufferedReader br = new BufferedReader(new InputStreamReader(System.in));
        int number = -1;
        List<String> expectedOutputList = FileUtils
                .readLines(new File("/home/sankalpkulshrestha/mnist/test-labels.csv"), "UTF-8");
        while ((number = Integer.parseInt(br.readLine())) != -1) {
            int output = neuralNetwork.test(new TrainingData.Sample(
                    new File("/home/sankalpkulshrestha/mnist/test-images/" + number + ".jpg"), 0));
            System.out.println(output);
        }
    }
}