Conditionally Strongly Log-Concave Generative Models

Abstract

There is a growing gap between the impressive results of deep image generative models and classical algorithms that offer theoretical guarantees. The former suffer from mode collapse or memorization issues, limiting their application to scientific data. The latter require restrictive assumptions such as log-concavity to escape the curse of dimensionality. We partially bridge this gap by introducing conditionally strongly log-concave (CSLC) models, which factorize the data distribution into a product of conditional probability distributions that are strongly log-concave. This factorization is obtained with orthogonal projectors adapted to the data distribution. It leads to efficient parameter estimation and sampling algorithms, with theoretical guarantees, although the data distribution is not globally log-concave. We show that several challenging multiscale processes are conditionally log-concave using wavelet packet orthogonal projectors. Numerical results are shown for physical fields such as the ${\phi }^4$ model and weak lensing convergence maps with higher resolution than in previous works.

arXiv comment28 pages, 12 figures, accepted at ICML 2023

References in corpus (21)

• Denoising Diffusion Probabilistic Models
• Weak Gravitational Lensing
• Hierarchical Text-Conditional Image Generation with CLIP Latents
• Diffusion Models Beat GANs on Image Synthesis
• Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
• Score-Based Generative Modeling through Stochastic Differential Equations
• Cascaded Diffusion Models for High Fidelity Image Generation
• Non-Gaussian information from weak lensing data via deep learning
• Extracting Training Data from Diffusion Models
• Random design analysis of ridge regression
• Weak lensing scattering transform: dark energy and neutrino mass sensitivity
• Exact minimax risk for linear least squares, and the lower tail of sample covariance matrices
• Do dark matter halos explain lensing peaks?
• Wavelet Flow: Fast Training of High Resolution Normalizing Flows
• Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions
• Fast Mixing of Multi-Scale Langevin Dynamics under the Manifold Hypothesis
• Statistical Efficiency of Score Matching: The View from Isoperimetry
• On Energy-Based Models with Overparametrized Shallow Neural Networks
• Improved Analysis of Score-based Generative Modeling: User-Friendly Bounds under Minimal Smoothness Assumptions
• Towards a Theory of Non-Log-Concave Sampling: First-Order Stationarity Guarantees for Langevin Monte Carlo
• Learning multi-scale local conditional probability models of images