问题
My friend and I was having a lot of trouble trying to implement the perceptron algorithm, but then I found this tutorial, it goes through a java implementation and then has some example code. I substituted my own data structures for there ones in the tutorial, and it works! :)
HOWEVER
I made this substitution in the most simplistic possible way, manually enumerating my data structures. This works as a proof of concept, for my experimental "toy" data, but most certainly is not able to tackle the real data I want to consider. It's far too rigid.
Perhaps someone more proficient in abstract thinking and loops would be able to show me how I can improve this code.
The important data structure to consider is Map<File, int[] >, it looks like this:
/data/test/sports/t.s_1.txt, [0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0]
/data/test/politics/t.p_0.txt, [0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0]
/data/test/atheism/t.a_0.txt, [0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
/data/test/science/t.s_0.txt, [1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 2, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0]
The code that needs to be generalized proceeds this sentence, you can also find the full programme on this github page.
Thank you for your consideration & happy Japanese/Chinese new year!
public static void perceptron( Set<String> GLOBO_DICT, Map<File, int[] > training_perceptron_input, Map<File, int[] > test_perceptron_input)
{
//number of features, number of x, y, z
int size_of_globo_dict = GLOBO_DICT.size();
//number of instances
int NUM_INSTANCES = training_perceptron_input.size();
//three variables (features) they enumerate by
//features, xyz, i also do that
double[] a00 = new double [NUM_INSTANCES];
double[] a01 = new double [NUM_INSTANCES];
double[] a02 = new double [NUM_INSTANCES];
double[] a03 = new double [NUM_INSTANCES];
double[] a04 = new double [NUM_INSTANCES];
double[] a05 = new double [NUM_INSTANCES];
double[] a06 = new double [NUM_INSTANCES];
double[] a07 = new double [NUM_INSTANCES];
double[] a08 = new double [NUM_INSTANCES];
double[] a09 = new double [NUM_INSTANCES];
double[] a10 = new double [NUM_INSTANCES];
double[] a11 = new double [NUM_INSTANCES];
double[] a12 = new double [NUM_INSTANCES];
double[] a13 = new double [NUM_INSTANCES];
double[] a14 = new double [NUM_INSTANCES];
double[] a15 = new double [NUM_INSTANCES];
double[] a16 = new double [NUM_INSTANCES];
double[] a17 = new double [NUM_INSTANCES];
double[] a18 = new double [NUM_INSTANCES];
double[] a19 = new double [NUM_INSTANCES];
double[] a20 = new double [NUM_INSTANCES];
double[] a21 = new double [NUM_INSTANCES];
double[] a22 = new double [NUM_INSTANCES];
double[] a23 = new double [NUM_INSTANCES];
double[] a24 = new double [NUM_INSTANCES];
double[] a25 = new double [NUM_INSTANCES];
double[] a26 = new double [NUM_INSTANCES];
double[] a27 = new double [NUM_INSTANCES];
double[] a28 = new double [NUM_INSTANCES];
double[] a29 = new double [NUM_INSTANCES];
double[] a30 = new double [NUM_INSTANCES];
//2d array for features
int x = 0;
int[][] feature_matrix = new int[ training_perceptron_input.size() ][ GLOBO_DICT.size() ];
String[][] output_label = new String[ training_perceptron_input.size() ][ 2 ];
for(Entry<File, int[]> entry : training_perceptron_input.entrySet())
{
int[] container =entry.getValue();
for(int j = 0; j < 4; j++)
{
feature_matrix[x][j] = container[j];
output_label[x][1] = entry.getKey().toString();
}
x++;
}
int[] outputs = new int [NUM_INSTANCES];
for(int g = 0; g < NUM_INSTANCES; g++)
{
a00[g] = feature_matrix[g][ 0];
a01[g] = feature_matrix[g][ 1];
a02[g] = feature_matrix[g][ 2];
a03[g] = feature_matrix[g][ 3];
a04[g] = feature_matrix[g][ 4];
a05[g] = feature_matrix[g][ 5];
a06[g] = feature_matrix[g][ 6];
a07[g] = feature_matrix[g][ 7];
a08[g] = feature_matrix[g][ 8];
a09[g] = feature_matrix[g][ 9];
a10[g] = feature_matrix[g][10];
a11[g] = feature_matrix[g][11];
a12[g] = feature_matrix[g][12];
a13[g] = feature_matrix[g][13];
a14[g] = feature_matrix[g][14];
a15[g] = feature_matrix[g][15];
a16[g] = feature_matrix[g][16];
a17[g] = feature_matrix[g][17];
a18[g] = feature_matrix[g][18];
a19[g] = feature_matrix[g][19];
a20[g] = feature_matrix[g][20];
a21[g] = feature_matrix[g][21];
a22[g] = feature_matrix[g][22];
a23[g] = feature_matrix[g][23];
a24[g] = feature_matrix[g][24];
a25[g] = feature_matrix[g][25];
a26[g] = feature_matrix[g][26];
a27[g] = feature_matrix[g][27];
a28[g] = feature_matrix[g][28];
a29[g] = feature_matrix[g][29];
a30[g] = feature_matrix[g][30];
if(output_label[g][1].equals( "/home/yamada/Workbench/SUTD/ISTD_50.570/assignments/practice_data/data/train/atheism/a_0.txt" ) )
{
outputs[g] = 1;
}
else
{
outputs[g] = 0;
}
}
double[] weights = new double[ GLOBO_DICT.size() + 1];// 32 for input variables and one for bias
double localError, globalError;
int i, p, iteration, output;
weights[ 0] = randomNumber(0,1);// w1
weights[ 1] = randomNumber(0,1);// w2
weights[ 2] = randomNumber(0,1);// w3
weights[ 3] = randomNumber(0,1);// w4
weights[ 4] = randomNumber(0,1);// w5
weights[ 5] = randomNumber(0,1);// w6
weights[ 6] = randomNumber(0,1);// w7
weights[ 7] = randomNumber(0,1);// w8
weights[ 8] = randomNumber(0,1);// w9
weights[ 9] = randomNumber(0,1);// w10
weights[10] = randomNumber(0,1);// w11
weights[11] = randomNumber(0,1);// w12
weights[12] = randomNumber(0,1);// w13
weights[13] = randomNumber(0,1);// w14
weights[14] = randomNumber(0,1);// w15
weights[15] = randomNumber(0,1);// w16
weights[16] = randomNumber(0,1);// w17
weights[17] = randomNumber(0,1);// w18
weights[18] = randomNumber(0,1);// w19
weights[19] = randomNumber(0,1);// w20
weights[20] = randomNumber(0,1);// w21
weights[21] = randomNumber(0,1);// w22
weights[22] = randomNumber(0,1);// w23
weights[23] = randomNumber(0,1);// w24
weights[24] = randomNumber(0,1);// w25
weights[25] = randomNumber(0,1);// w26
weights[26] = randomNumber(0,1);// w27
weights[27] = randomNumber(0,1);// w28
weights[28] = randomNumber(0,1);// w29
weights[29] = randomNumber(0,1);// w30
weights[30] = randomNumber(0,1);// w31
weights[31] = randomNumber(0,1);// this is the bias
iteration = 0;
do {
iteration++;
globalError = 0;
//loop through all instances (complete one epoch)
for (p = 0; p < NUM_INSTANCES; p++)
{
// calculate predicted class
output = calculateOutput(theta,
weights,
a00[p],
a01[p],
a02[p],
a03[p],
a04[p],
a05[p],
a06[p],
a07[p],
a08[p],
a09[p],
a10[p],
a11[p],
a12[p],
a13[p],
a14[p],
a15[p],
a16[p],
a17[p],
a18[p],
a19[p],
a20[p],
a21[p],
a22[p],
a23[p],
a24[p],
a25[p],
a26[p],
a27[p],
a28[p],
a29[p],
a30[p]);
// difference between predicted and actual class values
localError = outputs[p] - output;
//update weights and bias
weights[ 0] += LEARNING_RATE * localError * a00[p];
weights[ 1] += LEARNING_RATE * localError * a01[p];
weights[ 2] += LEARNING_RATE * localError * a02[p];
weights[ 3] += LEARNING_RATE * localError * a03[p];
weights[ 4] += LEARNING_RATE * localError * a04[p];
weights[ 5] += LEARNING_RATE * localError * a05[p];
weights[ 6] += LEARNING_RATE * localError * a06[p];
weights[ 7] += LEARNING_RATE * localError * a07[p];
weights[ 8] += LEARNING_RATE * localError * a08[p];
weights[ 9] += LEARNING_RATE * localError * a09[p];
weights[10] += LEARNING_RATE * localError * a10[p];
weights[11] += LEARNING_RATE * localError * a11[p];
weights[12] += LEARNING_RATE * localError * a12[p];
weights[13] += LEARNING_RATE * localError * a13[p];
weights[14] += LEARNING_RATE * localError * a14[p];
weights[15] += LEARNING_RATE * localError * a15[p];
weights[16] += LEARNING_RATE * localError * a16[p];
weights[17] += LEARNING_RATE * localError * a17[p];
weights[18] += LEARNING_RATE * localError * a18[p];
weights[19] += LEARNING_RATE * localError * a19[p];
weights[20] += LEARNING_RATE * localError * a20[p];
weights[21] += LEARNING_RATE * localError * a21[p];
weights[22] += LEARNING_RATE * localError * a22[p];
weights[23] += LEARNING_RATE * localError * a23[p];
weights[24] += LEARNING_RATE * localError * a24[p];
weights[25] += LEARNING_RATE * localError * a25[p];
weights[26] += LEARNING_RATE * localError * a26[p];
weights[27] += LEARNING_RATE * localError * a27[p];
weights[28] += LEARNING_RATE * localError * a28[p];
weights[29] += LEARNING_RATE * localError * a29[p];
weights[30] += LEARNING_RATE * localError * a30[p];
weights[31] += LEARNING_RATE * localError;
//summation of squared error (error value for all instances)
globalError += (localError*localError);
}
/* Root Mean Squared Error */
System.out.println("Iteration "+iteration+" : RMSE = "+Math.sqrt(globalError/NUM_INSTANCES));
}
while (globalError != 0 && iteration <= MAX_ITER);
System.out.println("\n=======\nDecision boundary equation:");
System.out.println(
" a00 *" +
weights[ 0] +
" a01 *" +
weights[ 1] +
" a02 *" +
weights[ 2] +
" a03 *" +
weights[ 3] +
" a04 *" +
weights[ 4] +
" a05 *" +
weights[ 5] +
" a06 *" +
weights[ 6] +
" a07 *" +
weights[ 7] +
" a08 *" +
weights[ 8] +
" a09 *" +
weights[ 9] +
" a10 *" +
weights[10] +
" a11 *" +
weights[11] +
" a12 *" +
weights[12] +
" a13 *" +
weights[13] +
" a14 *" +
weights[14] +
" a15 *" +
weights[15] +
" a16 *" +
weights[16] +
" a17 *" +
weights[17] +
" a18 *" +
weights[18] +
" a19 *" +
weights[19] +
" a20 *" +
weights[20] +
" a21 *" +
weights[21] +
" a22 *" +
weights[22] +
" a23 *" +
weights[23] +
" a24 *" +
weights[24] +
" a25 *" +
weights[25] +
" a26 *" +
weights[26] +
" a27 *" +
weights[27] +
" a28 *" +
weights[28] +
" a29 *" +
weights[29] +
" a30 *" +
weights[30] +
" bias: " +
weights[31]);
//2d array for features
int x_TEST = 0;
int[][] feature_matrix_TEST = new int[ test_perceptron_input.size() ][ GLOBO_DICT.size() ];
for(Entry<File, int[]> entry : test_perceptron_input.entrySet())
{
int[] container =entry.getValue();
for(int jj = 0; jj < 4; jj++)
{
feature_matrix_TEST[x_TEST][jj] = container[jj];
}
x_TEST++;
}
//three variables (features) they enumerate by
//features, xyz, i also do that
double[] z00 = new double [NUM_INSTANCES];
double[] z01 = new double [NUM_INSTANCES];
double[] z02 = new double [NUM_INSTANCES];
double[] z03 = new double [NUM_INSTANCES];
double[] z04 = new double [NUM_INSTANCES];
double[] z05 = new double [NUM_INSTANCES];
double[] z06 = new double [NUM_INSTANCES];
double[] z07 = new double [NUM_INSTANCES];
double[] z08 = new double [NUM_INSTANCES];
double[] z09 = new double [NUM_INSTANCES];
double[] z10 = new double [NUM_INSTANCES];
double[] z11 = new double [NUM_INSTANCES];
double[] z12 = new double [NUM_INSTANCES];
double[] z13 = new double [NUM_INSTANCES];
double[] z14 = new double [NUM_INSTANCES];
double[] z15 = new double [NUM_INSTANCES];
double[] z16 = new double [NUM_INSTANCES];
double[] z17 = new double [NUM_INSTANCES];
double[] z18 = new double [NUM_INSTANCES];
double[] z19 = new double [NUM_INSTANCES];
double[] z20 = new double [NUM_INSTANCES];
double[] z21 = new double [NUM_INSTANCES];
double[] z22 = new double [NUM_INSTANCES];
double[] z23 = new double [NUM_INSTANCES];
double[] z24 = new double [NUM_INSTANCES];
double[] z25 = new double [NUM_INSTANCES];
double[] z26 = new double [NUM_INSTANCES];
double[] z27 = new double [NUM_INSTANCES];
double[] z28 = new double [NUM_INSTANCES];
double[] z29 = new double [NUM_INSTANCES];
double[] z30 = new double [NUM_INSTANCES];
for(int g = 0; g < NUM_INSTANCES; g++)
{
z00[g] = feature_matrix_TEST[g][ 0];
z01[g] = feature_matrix_TEST[g][ 1];
z02[g] = feature_matrix_TEST[g][ 2];
z03[g] = feature_matrix_TEST[g][ 3];
z04[g] = feature_matrix_TEST[g][ 4];
z05[g] = feature_matrix_TEST[g][ 5];
z06[g] = feature_matrix_TEST[g][ 6];
z07[g] = feature_matrix_TEST[g][ 7];
z08[g] = feature_matrix_TEST[g][ 8];
z09[g] = feature_matrix_TEST[g][ 9];
z10[g] = feature_matrix_TEST[g][10];
z11[g] = feature_matrix_TEST[g][11];
z12[g] = feature_matrix_TEST[g][12];
z13[g] = feature_matrix_TEST[g][13];
z14[g] = feature_matrix_TEST[g][14];
z15[g] = feature_matrix_TEST[g][15];
z16[g] = feature_matrix_TEST[g][16];
z17[g] = feature_matrix_TEST[g][17];
z18[g] = feature_matrix_TEST[g][18];
z19[g] = feature_matrix_TEST[g][19];
z20[g] = feature_matrix_TEST[g][20];
z21[g] = feature_matrix_TEST[g][21];
z22[g] = feature_matrix_TEST[g][22];
z23[g] = feature_matrix_TEST[g][23];
z24[g] = feature_matrix_TEST[g][24];
z25[g] = feature_matrix_TEST[g][25];
z26[g] = feature_matrix_TEST[g][26];
z27[g] = feature_matrix_TEST[g][27];
z28[g] = feature_matrix_TEST[g][28];
z29[g] = feature_matrix_TEST[g][29];
z30[g] = feature_matrix_TEST[g][30];
}
int output_TEST;
// calculate predicted class TEST
output_TEST = calculateOutput(theta,
weights,
z00[2],
z01[2],
z02[2],
z03[2],
z04[2],
z05[2],
z06[2],
z07[2],
z08[2],
z09[2],
z10[2],
z11[2],
z12[2],
z13[2],
z14[2],
z15[2],
z16[2],
z17[2],
z18[2],
z19[2],
z20[2],
z21[2],
z22[2],
z23[2],
z24[2],
z25[2],
z26[2],
z27[2],
z28[2],
z29[2],
z30[2]);
System.out.println("\n=======\nTEST Point:");
System.out.println(
"z00[0]:" + z00[0] +
"z01[0]:" + z01[0] +
"z02[0]:" + z02[0] +
"z03[0]:" + z03[0] +
"z04[0]:" + z04[0] +
"z05[0]:" + z05[0] +
"z06[0]:" + z06[0] +
"z07[0]:" + z07[0] +
"z08[0]:" + z08[0] +
"z09[0]:" + z09[0] +
"z10[0]:" + z10[0] +
"z11[0]:" + z11[0] +
"z12[0]:" + z12[0] +
"z13[0]:" + z13[0] +
"z14[0]:" + z14[0] +
"z15[0]:" + z15[0] +
"z16[0]:" + z16[0] +
"z17[0]:" + z17[0] +
"z18[0]:" + z18[0] +
"z19[0]:" + z19[0] +
"z20[0]:" + z20[0] +
"z21[0]:" + z21[0] +
"z22[0]:" + z22[0] +
"z23[0]:" + z23[0] +
"z24[0]:" + z24[0] +
"z25[0]:" + z25[0] +
"z26[0]:" + z26[0] +
"z27[0]:" + z27[0] +
"z28[0]:" + z28[0] +
"z29[0]:" + z29[0] +
"z30[0]:" + z30[0]
);
System.out.println("class = "+output_TEST);
}
//end main
/**
* returns a random double value within a given range
* @param min the minimum value of the required range (int)
* @param max the maximum value of the required range (int)
* @return a random double value between min and max
*/
public static double randomNumber(int min , int max) {
DecimalFormat df = new DecimalFormat("#.####");
double d = min + Math.random() * (max - min);
String s = df.format(d);
double x = Double.parseDouble(s);
return x;
}
/**
* returns either 1 or 0 using a threshold function
* theta is 0range
* @param theta an integer value for the threshold
* @param weights[] the array of weights
* @param x the x input value
* @param y the y input value
* @param z the z input value
* @return 1 or 0
*/
static int calculateOutput(int theta,
double weights[],
double a00,
double a01,
double a02,
double a03,
double a04,
double a05,
double a06,
double a07,
double a08,
double a09,
double a10,
double a11,
double a12,
double a13,
double a14,
double a15,
double a16,
double a17,
double a18,
double a19,
double a20,
double a21,
double a22,
double a23,
double a24,
double a25,
double a26,
double a27,
double a28,
double a29,
double a30)
{
double sum = a00 *
weights[ 0] +
a01 *
weights[ 1] +
a02 *
weights[ 2] +
a03 *
weights[ 3] +
a04 *
weights[ 4] +
a05 *
weights[ 5] +
a06 *
weights[ 6] +
a07 *
weights[ 7] +
a08 *
weights[ 8] +
a09 *
weights[ 9] +
a10 *
weights[10] +
a11 *
weights[11] +
a12 *
weights[12] +
a13 *
weights[13] +
a14 *
weights[14] +
a15 *
weights[15] +
a16 *
weights[16] +
a17 *
weights[17] +
a18 *
weights[18] +
a19 *
weights[19] +
a20 *
weights[20] +
a21 *
weights[21] +
a22 *
weights[22] +
a23 *
weights[23] +
a24 *
weights[24] +
a25 *
weights[25] +
a26 *
weights[26] +
a27 *
weights[27] +
a28 *
weights[28] +
a29 *
weights[29] +
a30 *
weights[30] +
weights[31];
return (sum >= theta) ? 1 : 0;
}
}
回答1:
Here is a more generalized version your Perceptron class, simplified to reduce code duplication. Though untested, this is mechanically equivalent to the code you presented in your question.
class Perceptron {
static final int MAX_ITER = 100;
static final double LEARNING_RATE = 0.1;
static final int THETA = 0;
static final String FILEPATH =
"/home/yamada/Workbench/SUTD/ISTD_50.570/assignments/practice_data/data/train/atheism/a_0.txt";
public static void perceptron(Set<String> globoDict,
Map<File, int[]> trainingPerceptronInput,
Map<File, int[]> testPerceptronInput)
{
final int globoDictSize = globoDict.size(); // number of features (x, y, z)
// weights total 32 (31 for input variables and one for bias)
double[] weights = new double[globoDictSize + 1];
for (int i = 0; i < weights.length; i++) {
weights[i] = Math.floor(Math.random() * 10000) / 10000;
}
int inputSize = trainingPerceptronInput.size();
int[] outputs = new int[inputSize];
final double[][] a =
initializeOutput(trainingPerceptronInput, globoDictSize, outputs);
double globalError;
int iteration = 0;
do {
iteration++;
globalError = 0;
// loop through all instances (complete one epoch)
for (int p = 0; p < inputSize; p++) {
// calculate predicted class
int output = calculateOutput(THETA, weights, a, p);
// difference between predicted and actual class values
double localError = outputs[p] - output;
int i;
for (i = 0; i < a.length; i++) {
weights[i] += LEARNING_RATE * localError * a[i][p];
}
weights[i] += LEARNING_RATE * localError;
// summation of squared error (error value for all instances)
globalError += localError * localError;
}
/* Root Mean Squared Error */
System.out.println("Iteration "
+ iteration + " : RMSE = " + Math.sqrt(globalError
/ inputSize));
} while (globalError != 0 && iteration <= MAX_ITER);
System.out.println("\n=======\nDecision boundary equation:");
int i;
for (i = 0; i < a.length; i++) {
System.out.print(" a");
if (i < 10) System.out.print(0);
System.out.print(i + " *" + weights[i]);
}
System.out.println(" bias: " + weights[i]);
inputSize = testPerceptronInput.size();
outputs = new int[inputSize];
double[][] z = initializeOutput(testPerceptronInput, globoDictSize, outputs);
// calculate predicted class TEST
int output = calculateOutput(THETA, weights, z, 2);
System.out.println("\n=======\nTEST Point:");
for (i = 0; i < z.length; i++) {
System.out.print(" z");
if (i < 10) System.out.print(0);
System.out.print(i + "[0]" + z[i][0]);
}
System.out.println();
System.out.println("class = " + output);
}
static double[][] initializeOutput(
Map<File, int[]> perceptronInput, int size, int[] outputs)
{
final int inputSize = perceptronInput.size();
final double[][] a = new double[size][inputSize];
// 2d array for features
int[][] feature_matrix = new int[inputSize][size];
String[] output_label = new String[inputSize];
int x = 0;
for (Entry<File, int[]> entry : perceptronInput.entrySet()) {
int[] container = entry.getValue();
for (int j = 0; j < container.length; j++) {
feature_matrix[x][j] = container[j];
output_label[x] = String.valueOf(entry.getKey());
}
x++;
}
for (x = 0; x < inputSize; x++) {
for (int i = 0; i < a.length; i++) {
a[i][x] = feature_matrix[x][i];
}
outputs[x] = output_label[x].equals(FILEPATH) ? 1 : 0;
}
return a;
}
/**
* returns either 1 or 0 using a threshold function
* theta is 0range
* @param theta an integer value for the threshold
* @param weights the array of weights
* @param a the array of inputs
* @return 1 or 0
*/
static int calculateOutput(int theta, double[] weights, double[][] a, int index)
{
double sum = 0;
int i;
for (i = 0; i < a.length; i++) {
sum += weights[i] * a[i][index];
}
sum += weights[i];
return (sum >= theta) ? 1 : 0;
}
}
I hope this is instructional for you.
来源：https://stackoverflow.com/questions/28585880/manually-enumerated-data-structs-need-loops-to-generalize-perceptron-algorithm