https://github.com/prathmesh-ka-github/inkwell

Handwritten digit recognition engine. This project implements a neural network model to recognize handwritten digits from the MNIST dataset.
https://github.com/prathmesh-ka-github/inkwell

ai machine-learning mnist neural-networks

Last synced: 4 months ago
JSON representation

Handwritten digit recognition engine. This project implements a neural network model to recognize handwritten digits from the MNIST dataset.

Host: GitHub
URL: https://github.com/prathmesh-ka-github/inkwell
Owner: prathmesh-ka-github
Created: 2025-08-01T01:31:54.000Z (6 months ago)
Default Branch: main
Last Pushed: 2025-08-03T21:05:09.000Z (6 months ago)
Last Synced: 2025-08-25T23:13:50.734Z (6 months ago)
Topics: ai, machine-learning, mnist, neural-networks
Language: Jupyter Notebook
Homepage:
Size: 1.36 MB
Stars: 1
Watchers: 0
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

Inkwell

Handwritten digit recognition engine.

![neural networks](outputs/neuralnetworks.png)

This project implements a neural network model to recognize handwritten digits from the MNIST dataset. The MNIST dataset consists of 70,000 grayscale images of handwritten digits (0–9), each of size 28x28 pixels, split into 60,000 training images and 10,000 test images.

#### The model is trained using a feedforward neural network architecture with the following key components:

- Input Layer: Accepts 784 features (flattened 28x28 image)
- Hidden Layers: 2 densly connected layers with 128 and 64 neurons respectively with ReLU activation
- Output Layer: 10 neurons representing digit classes (0–9), using softmax activation

![model](outputs/20250802_21h09m57s_grim.png)

### Training Details
- Loss Function: Categorical Cross-Entropy
- Optimizer: Adam
- Learning Algorithm: Trained using backpropagation
- Batch Size: 1000
- Epochs: 5

![training](outputs/20250802_21h08m13s_grim.png)

### Results

The model achieves high accuracy on the test set and effectively generalizes to unseen handwritten digits. Evaluation metrics such as accuracy and loss can be visualized through training plots

![results](outputs/20250802_21h10m16s_grim.png)