A variational autoencoder (VAE) is a generative model that combines deep learning with Bayesian inference to learn compact latent representations of data. VAEs are widely used for image generation, anomaly detection, and data augmentation.
Fraud detection
TerrainGAN: a combination of VAE and GAN that can generate realistic 3D terrain
Flux.2-Klein-Edit-Ultra-Fast (Flux.2-Klein-Small-Decoder-Only) is a high-performance image editing and generation platform powered by the black-forest-labs/FLUX.2-klein-4B model paired with the black-forest-labs/FLUX.2-small-decoder VAE.
DEEP_NEPHRO is a university-led project focused on enhancing dialysis care through intelligent monitoring and optimization. By integrating machine learning and real-time data analysis, the project aims
Autoencoders, VAE, DCGAN on Pokémon dataset, and WGAN on MNIST — generative models built from scratch in PyTorch with latent space analysis and training progression visualizations
Functioning of neural network on the cifar 10 dataset
A physically interpretable, factored VAE for Gymnasium Lunar Lander: the lander's pose lives in named latent dims (x, y, cos/sin tilt) and the rest is a scene latent. Reproducible 3-stage training, shipped weights, a written report, and a self-contained interactive demo.