ProjNN-Ref.cpp

/**
 * @file ProjNN-ref.cpp
 * @brief Main application for data (spam) classification system
 *
 * Implements complete machine learning pipeline:
 * 1. Data loading and preprocessing
 * 2. Neural network training
 * 3. Interactive prediction interface
 */
#include <iostream>
#include "Matrix.h"
#include "NeuralNetwork.h"
#include "Processing.h"
#include "IDataset.h"
#include "Dataset.h"
int main() {
	// Polymorphic dataset
	IDataset* data = new Dataset();
	std::cout << "Set path to dataset: ";
	std::string path; std::cin >> path;
	data->load(path);

	// Build input matrix
	Processing preprocessor;
	Matrix<double> inputs = preprocessor.textToMatrix(data->getMessages());

	// Build targets matrix
	std::vector<std::vector<double>> targetRows;
	for (int i = 0; i < data->getSize(); i++) {
		double label = data->getLabel(i) ? 1.0 : 0.0;
		targetRows.push_back({ label });
	}
	Matrix<double> targets(targetRows);

	// Create and train network
	std::cout << "\033[33mTraining Neural Network.\033[0m" << std::endl;
	NeuralNetwork nn(inputs.getCols(), 16, 1, 0.01);
	nn.train(inputs, targets, 1000);
	std::cout << "\033[32mTraining Ended.\033[0m" << std::endl;

	// Prediction logic
	std::cout << "\033[33mSet Neural Network prediction mode:\033[0m" << std::endl;
	std::cout << "\033[34m1)\033[0m Single Input" << std::endl;
	std::cout << "\033[34m2)\033[0m Multiple from file" << std::endl;
	int mode; std::cout << "Mode: "; std::cin >> mode;

	while (mode < 1 || mode > 2) {
		std::cout << "\033[31mInvalid mode. \033[0m \nSet mode: "; std::cin >> mode;
	}

	if (mode == 1) {
		std::string input_text;
		std::cout << "Enter a message: ";
		std::cin.ignore();
		std::getline(std::cin, input_text);

		Matrix<double> test_input = preprocessor.textToMatrix(input_text);
		Matrix<double> prediction = nn.forward(test_input);
		std::cout << "Spam probability: \033[33m"
			<< prediction.getValues()[0][0] << "\033[0m\n";
	}
	else if (mode == 2) {
		std::string input_path;
		std::cout << "Set path to dataset: ";
		std::cin >> input_path;

		Dataset prediction_data;
		prediction_data.load(input_path);

		for (int i = 0; i < prediction_data.getSize(); i++) {
			Matrix<double> test_input = preprocessor.textToMatrix(
				prediction_data.getMessage(i)
			);
			Matrix<double> prediction = nn.forward(test_input);
			prediction_data.setLabel(i, prediction.getValues()[0][0] > 0.5);
		}
		prediction_data.save(input_path + ".out.csv");
		std::cout << "\033[32mPredictions saved to: "
			<< input_path << ".out.txt\033[0m" << std::endl;
	}

	// Cleanup
	delete data;
	return 0;
}