CoCoGen: Physically-Consistent and Conditioned Score-based Generative Models for Forward and Inverse Problems.
Christian JacobsenYilin ZhuangKarthik DuraisamyPublished in: CoRR (2023)
Keyphrases
- generative model
- inverse problems
- image reconstruction
- global optimization
- convex optimization
- probabilistic model
- discriminative models
- optimization problems
- optimization methods
- prior knowledge
- semi supervised
- hierarchical hidden markov models
- discriminative learning
- partial differential equations
- conditional random fields
- generative and discriminative models
- early vision
- object categories
- em algorithm
- smoothness constraint
- expectation maximization
- computationally expensive
- parameter space
- information extraction
- multiscale
- discriminative methods
- image segmentation