Handwritten-Digit-Classification-Using-ANN

A neural network built from scratch to classify handwritten digits from the MNIST dataset.

• Technologies Used: numpy
• Highlights:
  • Custom architecture with 3 dense layers
  • Achieved 97% accuracy
  • Confusion matrix and prediction examples

1. Model Architecture

2. Output

We assume the image data of 28x28 is flatern into 784x1 dimension: