{"id":"arxiv/2610.34484","url":"https://arxiv.org/abs/2610.34484","tags":["stat.ML","cs.LG","math.ST"],"score":0,"title":"Intrinsic Gradient Barriers in Pushforward Deep Generative Models","domain":"stat.ML","source":"arxiv","authors":["Edric Tam","Qingyang Liu","Barbara Engelhardt","Ray Bai"],"reviews":0,"subject":"stat.ML","summary":"Broad classes of deep generative models (DGMs) operate by learning to map a simple source distribution, such as a standard Gaussian, to a complex high-dimensional target distribution. Despite their success, DGMs often empirically struggle to accurately represent distributions with heavy tails or well-separated modes. In this work, we show that these difficulties reflect intrinsic properties of the underlying transport problem rather than particular neural network architectures or training procedures. We establish quantitative lower bounds on the gradients and Jacobians of pushforward…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-10-25","reactions":{}}