我有问题。我在C中找到了Artifical Neural Network代码，我想把它改成JAVA。所以我在功能上做了这些修改：
在C：
void ComputeFeedForwardSignals(double* MAT_INOUT, double* V_IN,double* V_OUT, double* V_BIAS,int size1,int size2,int layer)
{
int row,col;
for(row=0;row < size2; row++)
{
V_OUT[row]=0.0;
for(col=0;col<size1;col++)V_OUT[row]+=(*(MAT_INOUT+(row*size1)+col)*V_IN[col]);
V_OUT[row]+=V_BIAS[row];
if(layer==0) V_OUT[row] = tanh(V_OUT[row]);
}
}
在JAVA中使用相同的代码：
private void ComputeFeedForwardSignals (double[][] MAT_INOUT, double[] V_IN, double[] V_OUT, double[] V_BIAS, int size1, int size2, int layer) {
int row,col;
for(row=0;row < size2; row++)
{
V_OUT[row]=0.0;
for(col=0;col<size1;col++)V_OUT[row]+=(*(MAT_INOUT+(row*size1)+col)*V_IN[col]);
V_OUT[row]+=V_BIAS[row];
if(layer==0) V_OUT[row] = Math.tanh(V_OUT[row]);
}
}
...但我不知道，我怎么能取代这一行：
for(col=0;col<size1;col++)V_OUT[row]+=(*(MAT_INOUT+(row*size1)+col)*V_IN[col]);
请帮帮我。
...编辑
所以，这是原始的，完整的和有效的C代码（我只剪切大数组）：
//Analysis Type - Classification
#include <stdio.h>
#include <conio.h>
#include <math.h>
#include <stdlib.h>
double input_hidden_weights[100][320]=
{
{5.80887333084651e-001, 1.30476168251902e+000, 5.80288623607794e-001, 8.14671389077252e-001, 5.43029117736068e-001, 5.15946547751079e-001, 4.86324144066176e-001, 3.26116870507742e-001, 4.02847954975450e-001, 1.66380273429940e-001, 3.39504086983093e-001, 5.26249449907226e-002, 2.56145034309448e-001, -9.70569690724137e-002, 1.59242465161706e-001, -2.33214192739307e-001, 1.87648652577582e-002, -3.34041203799010e-001, -1.11393479757180e-001, -3.73367046579334e-001, -2.17964256784897e-001, -4.00192132238685e-001, -2.81209945053773e-001, -3.76012204270845e-001, -3.10614624230684e-001, -3.15921671544855e-001, -2.81529041789346e-001, -2.93125194852085e-001, -2.34501635198919e-001, -2.53483381689582e-001, -1.48248548624413e-001, -2.39004623741234e-001, -3.76904804526471e-002, -2.41802215751481e-001, 9.19371048359130e-002, -2.45416703676273e-001, 2.10015748048206e-001, -2.69629916475520e-001, 3.06584661619282e-001, -2.55206735127512e-001, 3.40664514242493e-001, -2.31156686442331e-001, 3.48619703956421e-001, -1.70205811278937e-001, 2.58540184919054e-001, -9.02211837839767e-002, 1.58219383097399e-001, 2.27214780937834e-002, 4.83874056064942e-002, 1.12670862875882e-001, -7.03828317119573e-002, 2.29852689788459e-001, -1.77576747217408e-001, 3.39353599529732e-001, -2.85410462161884e-001, 4.32495256767594e-001, -3.92336294795819e-001, //...//, 2.05201376126660e+000, 3.66183884164477e-001, -3.13236249649404e-001, -6.90288932727299e-001, -4.03350305378540e-001, -5.18217246345780e-001, -2.11389978349476e-002 }
};
double output_bias[23]={ 9.26583664006470e-001, -5.12649304998720e-001, 8.50167833795463e-001, -3.12847753694448e+000, 3.03755707027426e+000, -2.37819656326885e+000, 2.64402241833182e+000, 2.54741061276513e+000, -1.52779355163716e+000, -1.24383441583121e+000, 3.72867458316797e+000, 4.15165486353238e+000, -4.88008803506142e+000, -3.35612469382435e+000, -2.74153025313899e+000, 3.11815950006976e+000, 1.49738419343937e+000, -3.43954471203446e+000, 3.03236240807163e-001, -4.77180501599233e+000, 3.08664451646140e+000, -1.66680993545569e-002, 2.12529133729690e+000 };
double max_input[320]={ 1.54200000000000e+003, //...//, 1.86000000000000e+002, 2.24000000000000e+002 };
double min_input[320]={ -1.72800000000000e+003, -1.92000000000000e+003, -2.50800000000000e+003, -2.95900000000000e+003, //...//, -1.13000000000000e+002, -1.01000000000000e+002, -1.19000000000000e+002, -9.30000000000000e+001, -1.18000000000000e+002, -1.01000000000000e+002 };
double input[320] = {618, 1067, 499, 1179, 358, 1187, //...// , 0, 0, 0, 0};
double hidden[100];
double output[23];
double MeanInputs[320]={ 1.24955952380952e+002, 5.69701190476191e+002, 1.37892261904762e+002, //...//, 3.82738095238095e-001, 6.29166666666667e-001 };
void FindMax(double* vec, double* max, long* maxIndex,int len)
{
long i;
*max = vec[0];
*maxIndex = 0;
for(i=1; i<len; i++)
{
if(vec[i]>*max)
{
*max = vec[i];
*maxIndex = i;
}
}
}
void ScaleInputs(double* input, double minimum, double maximum, int size)
{
double delta;
long i;
for(i=0; i<size; i++)
{
delta = (maximum-minimum)/(max_input[i]-min_input[i]);
input[i] = minimum - delta*min_input[i]+ delta*input[i];
}
}
void softmax(double* vec,int len)
{
long i, j;
double sum=0.0;
for(i=0; i<len; i++)
{
if(vec[i]>200)
{
double max;
long maxIndex;
FindMax(vec, &max, &maxIndex,len);
for(j=0; j<len; j++)
{ if(j==maxIndex) vec[j] = 1.0;
else vec[j] = 0.0;
}
return;
}
else
{
vec[i] = exp(vec[i]);
}
sum += vec[i];
}
if(sum==0)
{
long a = 1;
}
if(sum!=0.0)
{
for(i=0; i<len; i++) vec[i] = vec[i]/sum;
}
else for(i=0; i<len; i++) vec[i] = 1.0/(double)len;
}
void ComputeFeedForwardSignals(double* MAT_INOUT,double* V_IN,double* V_OUT, double* V_BIAS,int size1,int size2,int layer)
{
int row,col;
for(row=0;row < size2; row++)
{
V_OUT[row]=0.0;
for(col=0;col<size1;col++) V_OUT[row]+=(*(MAT_INOUT+(row*size1)+col)*V_IN[col]);
V_OUT[row]+=V_BIAS[row];
if(layer==0) V_OUT[row] = tanh(V_OUT[row]);
}
}
void RunNeuralNet_Classification ()
{
ComputeFeedForwardSignals((double*)input_hidden_weights,input,hidden,hidden_bias,320, 100,0);
ComputeFeedForwardSignals((double*)hidden_output_wts,hidden,output,output_bias,100, 23,1);
}
int main()
{
int cont_inps;
int index;
int i=0;
int keyin=1;
double max;
while(1)
{
max=3.e-300;
for(cont_inps=0;cont_inps<320;cont_inps++)
{
//Substitution of missing continuous variables
if(input[cont_inps] == -9999)
input[cont_inps]=MeanInputs[cont_inps];
}
ScaleInputs(input,0,1,320);
RunNeuralNet_Classification();
//Output Activation is Softmax;
softmax(output, 23);
for(i=0;i<23;i++)
{
if(max<output[i])
{
max=output[i];
index=i+1;
}
}
printf("\n%s","Predicted category = ");
switch(index)
{
case 1: printf("%s\n","A"); break;
case 2: printf("%s\n","B"); break;
case 3: printf("%s\n","C"); break;
case 4: printf("%s\n","D"); break;
case 5: printf("%s\n","E"); break;
case 6: printf("%s\n","F"); break;
case 7: printf("%s\n","G"); break;
case 8: printf("%s\n","H"); break;
case 9: printf("%s\n","I"); break;
case 10: printf("%s\n","J"); break;
case 11: printf("%s\n","K"); break;
case 12: printf("%s\n","L"); break;
case 13: printf("%s\n","M"); break;
case 14: printf("%s\n","N"); break;
case 15: printf("%s\n","O"); break;
case 16: printf("%s\n","P"); break;
case 17: printf("%s\n","R"); break;
case 18: printf("%s\n","S"); break;
case 19: printf("%s\n","T"); break;
case 20: printf("%s\n","U"); break;
case 21: printf("%s\n","W"); break;
case 22: printf("%s\n","Y"); break;
case 23: printf("%s\n","Z"); break;
default: break;
}
printf("\n%s%.14f","Confidence level = ",max);
printf("\n\n%s\n","Press any key to make another prediction or enter 0 to quit the program.");
keyin=getch();
if(keyin==48)break;
}
return 0;
}
JAVA中的代码效果不佳：
public class MAIN {
public static void main(String[] args) {
network Neurons = new network();
Neurons.Read();
Neurons.SSN();
}
}
import java.io.BufferedReader;
import java.io.FileNotFoundException;
import java.io.FileReader;
import java.io.IOException;
public class network {
private double[][] Input_hidden_weights = new double[100][320];
private double[] Hidden_bias = new double[100];
private double[][] Hidden_output_wts = new double[23][100];
private double[] Output_bias = new double[23];
private double[] Max_input = new double[320];
private double[] Min_input = new double[320];
private double[] Input = new double[320];
private double[] Hidden = new double[100];
private double[] Output = new double[23];
private double[] MeanInputs = new double[320];
public void Read() {
this.Read_Input_hidden_weights();
this.Hidden_bias();
this.Hidden_output_wts();
this.Output_bias();
this.Max_input();
this.Min_input();
this.Input();
this.MeanInputs();
}
private void Read_Input_hidden_weights() {
FileReader fr = null;
String linia = "";
int a=0;
int b=0;
try {
fr = new FileReader("Ihw.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Input_hidden_weights[a][b] = Double.parseDouble(linia);
b++;
if (b==320) {
b=0;
a++;
}
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Hidden_bias() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("Hb.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Hidden_bias[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Hidden_output_wts() {
FileReader fr = null;
String linia = "";
int a=0;
int b=0;
try {
fr = new FileReader("How.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Hidden_output_wts[a][b] = Double.parseDouble(linia);
b++;
if (b==100) {
b=0;
a++;
}
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Output_bias() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("Ob.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Output_bias[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Max_input() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("Mi.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Max_input[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Min_input() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("Mini.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Min_input[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void Input() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("I.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
Input[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
private void MeanInputs() {
FileReader fr = null;
String linia = "";
int a=0;
try {
fr = new FileReader("Mei.txt");
} catch (FileNotFoundException e) {
System.out.println("ERR");
System.exit(1);
}
BufferedReader bfr = new BufferedReader(fr);
try {
while((linia = bfr.readLine()) != null){
MeanInputs[a] = Double.parseDouble(linia);
a++;
}
} catch (IOException e) {
System.out.println("ERR");
System.exit(2);
}
try {
fr.close();
} catch (IOException e) {
System.out.println("ERR");
System.exit(3);
}
}
public void SSN() {
int cont_inps;
int index = 0;
int i = 0;
double max;
max=3.e-300;
for (cont_inps=0; cont_inps<320; cont_inps++) {
if (Input[cont_inps] == -9999) Input[cont_inps] = MeanInputs[cont_inps];
}
ScaleInputs(Input, 0, 1, 320);
RunNeuralNet_Classification();
//Output Activation is Softmax;
Softmax(Output, 23);
for(i=0; i<23; i++) {
if(max < Output[i]) {
max = Output[i];
index = i+1;
}
}
System.out.printf("Predicted category = ");
switch(index)
{
case 1: System.out.printf("A"); break;
case 2: System.out.printf("B"); break;
case 3: System.out.printf("C"); break;
case 4: System.out.printf("D"); break;
case 5: System.out.printf("E"); break;
case 6: System.out.printf("F"); break;
case 7: System.out.printf("G"); break;
case 8: System.out.printf("H"); break;
case 9: System.out.printf("I"); break;
case 10: System.out.printf("J"); break;
case 11: System.out.printf("K"); break;
case 12: System.out.printf("L"); break;
case 13: System.out.printf("M"); break;
case 14: System.out.printf("N"); break;
case 15: System.out.printf("O"); break;
case 16: System.out.printf("P"); break;
case 17: System.out.printf("R"); break;
case 18: System.out.printf("S"); break;
case 19: System.out.printf("T"); break;
case 20: System.out.printf("U"); break;
case 21: System.out.printf("W"); break;
case 22: System.out.printf("Y"); break;
case 23: System.out.printf("Z"); break;
default: break;
}
System.out.printf(", Confidence level = " + max);
}
void FindMax(double vec[], double max[], long maxIndex[], int len)
{
int i;
max[0] = vec[0];
maxIndex[0] = 0;
for(i=1; i<len; i++) {
if(vec[i] > max[0]) {
max[0] = vec[i];
maxIndex[0] = i;
}
}
}
private void ScaleInputs(double Input[], double minimum, double maximum, int size) {
double delta;
int i=0;
for (i=0; i<size; i++) {
delta = (maximum-minimum)/(this.Max_input[i] - this.Min_input[i]);
Input[i] = minimum - delta * this.Min_input[i] + delta * Input[i];
}
}
private void Softmax (double vec[], int len) {
int i, j;
double sum=0.0;
for(i=0; i<len; i++)
{
if(vec[i]>200)
{
double[] max = new double[1];
long[] maxIndex = new long[1];
FindMax(vec, max, maxIndex, len);
for(j=0; j<len; j++)
{ if(j==maxIndex[0]) vec[j] = 1.0;
else vec[j] = 0.0;
}
return;
}
else
{
vec[i] = Math.exp(vec[i]);
}
sum += vec[i];
}
if(sum==0)
{
//long a = 1;
}
if(sum!=0.0)
{
for(i=0; i<len; i++) vec[i] = vec[i]/sum;
}
else for(i=0; i<len; i++) vec[i] = 1.0/(double)len;
}
private void ComputeFeedForwardSignals (double[][] MAT_INOUT, double[] V_IN, double[] V_OUT, double[] V_BIAS, int size1, int size2, int layer) {
int row, col;
for(row=0;row < size2; row++)
{
V_OUT[row]=0.0;
for(col=0; col<size1; col++) V_OUT[row] =+ MAT_INOUT[row][col] * V_IN[col];
V_OUT[row]+=V_BIAS[row];
if(layer==0) V_OUT[row] = Math.tanh(V_OUT[row]);
}
}
private void RunNeuralNet_Classification()
{
ComputeFeedForwardSignals(Input_hidden_weights, Input, Hidden, Hidden_bias, 320, 100, 0);
ComputeFeedForwardSignals(Hidden_output_wts, Hidden, Output, Output_bias, 100, 23, 1);
}
}
在我正在打开的文件中，我放了所有数据，具有以下结构：
5.80887333084651e-001
1.30476168251902e+000
5.80288623607794e-001
8.14671389077252e-001
5.43029117736068e-001
5.15946547751079e-001
4.86324144066176e-001
3.26116870507742e-001
4.02847954975450e-001
1.66380273429940e-001
3.39504086983093e-001
5.26249449907226e-002
2.56145034309448e-001
..... etc.
这些是您可以在C代码中的数组中看到的值。什么是坏事？
答案 0 :(得分：3)
计算
V_OUT[row]+=(*(MAT_INOUT+(row*size1)+col)*V_IN[col]);
包含[row] [column] 2d数组索引对的手动转换为单维数组索引。这行可以写成
V_OUT[row] += MAT_INOUT[row * size1 + col] * V_IN[col];
我想这会更容易理解！
在（旧方言）C中需要这种代码来实现可变大小的2d数组。在Java中你需要
V_OUT[row] += MAT_INOUT[row][col] * V_IN[col];