🧠 MNIST Image Classifier using Multilayer Perceptron in Java

This project implements a simple image classifier for the MNIST dataset, using a Multilayer Perceptron (MLP) neural network written in Java

🚀 Features

• Classifies handwritten digits (0–9) from the MNIST dataset.
• Implemented using Multilayer Perceptron (Feedforward Neural Network).
• Includes training, testing, and evaluation modules.
• Optional model saving/loading.

---

🧱 Architecture

• Input Layer: 784 neurons (28x28 pixel grayscale images)
• Output Layer: 10 neurons (one for each digit class)
• Loss Function: Cross-Entropy
• Optimizer: Stochastic Gradient Descent (SGD)

---

📦 Dependencies

If you're using pure Java:

• Java 8 or above
• Maven or Gradle for build