Handcrafted Digit Recognizer with Numpy

This project is a custom implementation of a digit recognizer using only NumPy. It includes algorithms for gradient descent, backpropagation, and activation functions like ReLU and softmax. The project focuses on building a neural network from scratch in Python, without relying on external libraries for deep learning.

Dataset

This model uses the MNIST Dataset for training. It contains grayscale images of handwritten digits (0-9) and their corresponding labels.

Implementation and Usage

This project can be implemented locally by mentioning the dataset's exact location. The project consists of several code cells that perform the following tasks:

1. Data Preprocessing: Reading and preprocessing the dataset, including shuffling and normalization.
2. Initialization: Initializing the parameters (weights and biases) of the neural network.
3. Forward Propagation: Implementing forward propagation to compute the outputs of the neural network.
4. Backward Propagation: Implementing backward propagation to compute the gradients of the parameters.
5. Gradient Descent: Implementing gradient descent to update the parameters based on the gradients.
6. Prediction and Testing: Making predictions using the trained model and testing the accuracy on the training set.

Dependencies

• NumPy
• Pandas
• Matplotlib