#include <stdio.h>
#include <stdlib.h>
// 定义卷积神经网络的结构
typedef struct {
float** weights; // 权重矩阵
float* biases; // 偏置向量
int numFilters; // 过滤器数量
int filterSize; // 过滤器大小
int inputSize; // 输入大小
} ConvNet;
// 创建卷积神经网络模型
ConvNet* createConvNet(int numFilters, int filterSize, int inputSize) {
ConvNet* net = (ConvNet*)malloc(sizeof(ConvNet));
net->numFilters = numFilters;
net->filterSize = filterSize;
net->inputSize = inputSize;
// 初始化权重矩阵和偏置向量
net->weights = (float**)malloc(numFilters * sizeof(float*));
for (int i = 0; i < numFilters; i++) {
net->weights[i] = (float*)malloc(filterSize * filterSize * sizeof(float));
for (int j = 0; j < filterSize * filterSize; j++) {
net->weights[i][j] = (float)rand() / RAND_MAX; // 随机初始化权重
}
}
net->biases = (float*)malloc(numFilters * sizeof(float));
for (int i = 0; i < numFilters; i++) {
net->biases[i] = 0.0; // 初始化偏置为0
}
return net;
}
// 使用卷积神经网络进行前向传播
void forwardPropagation(ConvNet* net, float* input, float* output) {
int outputSize = net->inputSize - net->filterSize + 1; // 计算输出大小
for (int i = 0; i < net->numFilters; i++) {
for (int j = 0; j < outputSize; j++) {
for (int k = 0; k < outputSize; k++) {
float sum = 0.0;
for (int m = 0; m < net->filterSize; m++) {
for (int n = 0; n < net->filterSize; n++) {
sum += input[(j + m) * net->inputSize + (k + n)] * net->weights[i][m * net->filterSize + n];
}
}
output[i * outputSize * outputSize + j * outputSize + k] = sum + net->biases[i];
}
}
}
}
int main() {
// 创建一个卷积神经网络模型
int numFilters = 3;
int filterSize = 3;
int inputSize = 5;
ConvNet* net = createConvNet(numFilters, filterSize, inputSize);
// 输入数据
float input[inputSize * inputSize] = {0.5, 0.6, 0.7, 0.8, 0.9,
0.1, 0.2, 0.3, 0.4, 0.5,
0.6, 0.7, 0.8, 0.9, 1.0,
0.1, 0.2, 0.3, 0.4, 0.5};
// 计算输出
int outputSize = inputSize - filterSize + 1;
float output[numFilters * outputSize * outputSize];
forwardPropagation(net, input, output);
// 打印输出
printf("Output:\n");
for (int i = 0; i < numFilters; i++) {
printf("Filter %d:\n", i);
for (int j = 0; j < outputSize; j++) {
for (int k = 0; k < outputSize; k++) {
printf("%.2f ", output[i * outputSize * outputSize + j * outputSize + k]);
}
printf("\n");
}
printf("\n");
}
// 释放内存
for (int i = 0; i < numFilters; i++) {
free(net->weights[i]);
}
free(net->weights);
free(net->biases);
free(net);
return 0;
}
上述代码演示了一个简单的卷积神经网络模型,创建了一个包含3个过滤器(filters)的模型,每个过滤器的大小为3x3。然后给定一个5x5的输入矩阵,通过前向传播计算了输出矩阵。最后打印了输出矩阵。
请注意,此代码仅用于演示目的,实际的卷积神经网络模型可能需要更复杂的结构和算法来实现。