Objective: To implement a VAE, train it on a dataset of your choice (e.g., MNIST, Fashion MNIST, or a dataset of images), and generate new samples from the learned latent space. Dataset: Choose one of ...

Detect unknown (zero-day) IoT attacks using unsupervised anomaly detection with deep autoencoders. Train only on normal traffic - no attack labels needed!