TensorFlow implementation of the SOM-VAE model as described in https://arxiv.org/abs/1806.02199

🧠 Explore memory and stability in AI to prevent forgetting and enable modular learning through innovative neural network experiments.

🔤 Master three-word sentences for clear English communication through simple SV and SVO structures. Explore examples and enhance your language skills.

🔍 Evaluate synthetic data quality against real tabular datasets with Autocurator, measuring fidelity, coverage, privacy, and utility through clear metrics and visual reports.

🛠️ Develop a Variational Autoencoder to generate realistic tabular data, capturing complex patterns and relationships for data analysis and modeling.

🧑🎨 Generate and compare synthetic tabular data using Gaussian Copula and Variational Autoencoders for enhanced analytics and model prototyping.

🌱 Benchmark embedding methods for survival prediction on omics data, comparing classical and deep-learning approaches for improved analysis and insights.

🔍 Measure data authenticity and quality in synthetic analytics for safer AI. Explore relationships, diversity, and truthfulness in modern machine learning.

VAEs with PyTorch + Lightning

A comprehensive collection of implementations for Deep Generative Models, including VAEs, Normalizing Flows (MAF), CycleGAN, Energy-Based Models (EBM), and Score-Based Generative Models (NCSN). Features applications in Anomaly Detection, Disentanglement, and Style Transfer.

Centralized terminal for Open Source AI Resources (Checkpoints, Encoders, VAEs). Integrated with KSimply database.

Semi-supervised deep learning pipeline for SETI (Search for Extra-Terrestrial Intelligence) program. Uses custom β-VAE + Random Forest to detect narrowband signals with Doppler drift in Radio Telescope observations.

Advanced Deep Learning with Keras, published by Packt

Advanced VAE-based pipeline for automated cell type annotation in single-cell RNA-seq data with Optuna optimization and 99%+ accuracy

A research-focused modular generative modeling library built on JAX/Flax NNX

Deep reinforcement learning for DIY self-driving cars. Train JetBot, JetRacer, or DonkeySim to drive in minutes on edge (Jetson Nano) or in simulation.

VAE, GAN, diffusion models for synthetic data generation and augmentation

Lecture slides for the Course "Large Foundation Models: Mathematics, Algorithms, and Applications"

High-complexity implementation of Probabilistic Models (VAE & Normalizing Flows). Features dual-mode manifold discovery through $\beta$-regularization and Evidence Lower Bound (ELBO) maximization. Validated via FID (0.4473), utilizing TFP Bijectors and MultivariateNormalTriL layers.

Experiments with VAEs on MNIST dataset using Flux.jl