Neural Network for MNIST Digit Recognition: Multi-Class Classification of Handwritten Digits

Overview

This project involves training a neural network model to classify handwritten digits from the MNIST dataset using TensorFlow and Keras.

Model Architecture

The model's architecture is as follows:

1. Input Layer:

   • Flattens the input image dimensions of 28x28 pixels to a 1D array.

2. First Dense Layer:

   • Units: 128
   • Activation: ReLU
   • Kernel Size: 784x128
   • Bias Size: 128

3. Second Dense Layer:

   • Units: 32
   • Activation: ReLU
   • Kernel Size: 128x32
   • Bias Size: 32

4. Output Dense Layer:

   • Units: 10 (one for each digit)
   • Activation: Softmax
   • Kernel Size: 32x10
   • Bias Size: 10

Dependencies

• Python 3.8 or higher
• TensorFlow 2.x
• NumPy
• Matplotlib (for plotting training results)

Installation

To set up the project environment:

pip install tensorflow numpy matplotlib