Two talks will be presented, with a 5-minute intermission (1:35-1:40) in between. Feel free to come to one, the other, or both.  Hamidreza will speak second. 

Abstract:  We present a comprehensive examination of score-based diffusion models of AI for generating synthetic images. These models hinge upon a dynamic auxiliary time mechanism driven by stochastic differential equations, wherein the score function is acquired from input images. Our investigation unveils a criterion for evaluating efficiency of the score-based diffusion models: the power of the generative process depends on the ability to de-construct fast correlations during the reverse/de-noising phase. To improve the quality of the produced synthetic images, we introduce an approach coined "U-Turn Diffusion". In this talk we investigate its properties in detail.