Physicist in my past life, generative modeling in my current!
We propose optimal likelihood ratio-based selective classification methods based on the Neyman-Pearson lemma and evaluate them under vision and language covariate shifts tasks.
We show that a single unconditional diffusion model performs competitively in out-of-distribution detection tasks by measuring the rate-of-change and curvature of diffusion paths connecting data samples to the standard normal distribution.
We extend gradient flow methods to a variety of high-quality image synthesis tasks using a novel density ratio learning method.
We apply techniques from continual learning to the problem of selective forgetting in deep generative models. Our method, dubbed Selective Amnesia, allows users to remap undesired concepts to user-defined ones.
