A0155
Title: Statistical generative learning leveraging pretrained large models
Authors: Jian Huang - The Hong Kong Polytechnic University (China) [presenting]
Abstract: The focus is on statistical generative models leveraging pre-trained large models. The basics of two generative learning approaches are first introduced: generative adversarial networks and denoising diffusion models. Two examples are then used to illustrate generative modeling with the help of a pre-trained large model. The first example considers protein data analysis based on data representations learned through a pre-trained model using a protein sequence database. The second example demonstrates a Bayesian fine-tuning approach for image generation leveraging a pretrained diffusion model.
