main.cpp

#include "NeuralNetwork.h"

#include <cmath>
#include <iomanip>
#include <iostream>
#include <string>
#include <vector>

int main() {
    // 读取配置并训练模型
    std::cout << "简单神经网络回归演示" << std::endl;

    int hiddenLayerCount = 0;
    std::cout << "请输入隐藏层数量（0-3）：";
    std::cin >> hiddenLayerCount;
    if (hiddenLayerCount < 0) {
        hiddenLayerCount = 0;
    }
    if (hiddenLayerCount > 3) {
        hiddenLayerCount = 3;
    }

    std::vector<int> hiddenLayers;
    for (int i = 0; i < hiddenLayerCount; ++i) {
        int neurons = 1;
        std::cout << "请输入第 " << (i + 1) << " 个隐藏层的神经元数量（1-10）：";
        std::cin >> neurons;
        if (neurons < 1) {
            neurons = 1;
        }
        if (neurons > 10) {
            neurons = 10;
        }
        hiddenLayers.push_back(neurons);
    }

    double learningRate = 0.01;
    std::cout << "请输入学习率：";
    std::cin >> learningRate;
    if (learningRate <= 0.0) {
        learningRate = 0.01;
    }

    int epochs = 200;
    std::cout << "请输入训练轮次（>=200）：";
    std::cin >> epochs;
    if (epochs < 200) {
        epochs = 200;
    }

    // 创建神经网络对象
    NeuralNetwork network(hiddenLayers, learningRate);

    std::string filename;
    std::cout << "请输入数据文件路径：";
    std::cin >> filename;

    if (!network.loadData(filename)) {
        std::cout << "读取数据文件失败。" << std::endl;
        return 1;
    }

    // 训练并输出参数
    network.train(epochs);
    network.printParameters();

    // 预测与误差输出
    double inputX = 0.0;
    std::cout << "请输入新的输入 x：";
    std::cin >> inputX;
    double predictedY = network.predict(inputX);
    std::cout << "预测的 y 值为：" << std::fixed << std::setprecision(6) << predictedY << std::endl;

    double realY = 0.0;
    std::cout << "请输入真实的 y 值：";
    std::cin >> realY;
    double error = std::fabs(predictedY - realY);
    std::cout << "绝对误差为：" << std::fixed << std::setprecision(6) << error << std::endl;

    return 0;
}