This repository contains a Java implementation of a Multilayer Perceptron (MLP) neural network for solving a supervised discrete classification problem (SDT).
The implementation supports multi-class classification using a fully connected neural network with three hidden layers and configurable activation functions. The model is trained using mini-batch gradient descent and Mean Squared Error (MSE) loss.
The project uses separate training and testing datasets (train.txt and test.txt) in TXT format and includes functionality for training, evaluation and prediction.
A detailed explanation of the MLP implementation, training procedure and SDT classification problem is included in:
Report.pdf
The first page of the report lists the SDT classes used in the project.
train.txt– dataset used for training the MLPtest.txt– dataset used for evaluating the MLP
Both files must follow the format: x1,x2,label (no header).
- Training the network with mini-batch gradient descent
- Predicting labels for new samples
- Evaluating accuracy on the test set