This project focuses on building and evaluating autoencoders (AEs) and variational autoencoders (VAEs) for image generation and anomaly detection using the MNIST dataset.

• Goal: Train autoencoders and VAEs to generate images and detect anomalies.
• Data: MNIST dataset (both monochrome and color versions), used for training and testing.
• Models: Standard autoencoders for reconstructing images, and variational autoencoders for probabilistic generation.

• Generative Models: Train AEs and VAEs to generate new images based on learned data distributions.
• Anomaly Detection: Detect anomalies in images by evaluating reconstruction errors.