Generative Neural Network
Generative neural networks are artificial intelligence models designed to create new data instances that resemble a training dataset, achieving this by learning the underlying data distribution. Current research focuses on improving the quality and efficiency of generation across diverse applications, employing architectures like GANs, VAEs, diffusion models, and transformers, often tailored to specific data types (images, music, tabular data, etc.). These advancements are significantly impacting various fields, enabling applications such as image enhancement, fast simulations in high-energy physics, synthetic data generation for improved machine learning model training, and even novel approaches to scientific modeling and optimization.
Papers
Fast refacing of MR images with a generative neural network lowers re-identification risk and preserves volumetric consistency
Nataliia Molchanova, Bénédicte Maréchal, Jean-Philippe Thiran, Tobias Kober, Till Huelnhagen, Jonas Richiardi
Neural Task Synthesis for Visual Programming
Victor-Alexandru Pădurean, Georgios Tzannetos, Adish Singla