This project implements an autoencoder neural network to learn compact representations of handwritten digit images. Built and trained using Google Colab, the model encodes input images into a lower-dimensional latent space and reconstructs them with minimal information loss.

The autoencoder is trained on grayscale handwritten digit datasets and demonstrates capabilities such as image reconstruction, denoising, and feature learning. This project serves as a practical introduction to unsupervised learning and deep learning workflows in a cloud-based environment.

Features:

Encoder-decoder architecture for image compression and reconstruction Training and experimentation in Google Colab (no local setup required) Latent space representation learning Image denoising capability Visualization of original vs reconstructed images

Tech Stack:

Python TensorFlow / PyTorch Google Colab NumPy, Matplotlib

Use Cases:

Understanding autoencoders and representation learning Image denoising and reconstruction Dimensionality reduction Learning deep learning workflows using cloud notebooks