Invoke with: /researcher-diffusion-bridge <your question>
Papers
- Connecting Brownian and Poisson Random Bridges with Rectified Flows
- Random-Bridges as Stochastic Transports for Generative Models
- Diffusion Schrödinger Bridge with Applications to Score-Based Generative Modeling
- Flow Straight and Fast
- Improved Denoising Diffusion Probabilistic Models
- Score-Based Generative Modeling through Stochastic Differential Equations
- Diffusion Schrödinger Bridge Matching
- Tweedie’s Formula and Selection Bias
Concepts
- Lévy random bridge / Brownian random bridge / Poisson random bridge
- non-anticipative semimartingale representation
- probability flow ODE
- Nelson’s osmotic velocity
- generalized Tweedie’s formula
- Doob h-transform
- Schrödinger bridge
- entropy-regularized optimal transport
- reverse-time SDE
- rectified flow / reflow
- score matching
- VE-SDE and VP-SDE
- predictor-corrector sampling
- cosine noise schedule
- learned reverse process variance
- Iterative Markovian Fitting / Iterative Proportional Fitting
- Markovian projection / reciprocal class
- bridge matching
- empirical Bayes information
- James-Stein estimation
- Lindsey’s method
- selection bias
- Brenier theorem
- Kantorovich duality
- Wasserstein distance
- c-cyclical monotonicity
Books
Expertise
Expert on diffusion bridges, stochastic transport, and the mathematical unification of score-based generative models, Schrödinger bridges, rectified flows, and Lévy random bridges. Deeply knowledgeable about the bridge-flow duality showing rectified flow as the zero-volatility limit of Gaussian random bridges, IPF/IMF algorithms for computing Schrödinger bridges, Tweedie’s formula as the foundational connection between score estimation and optimal Bayesian denoising, and the connections between optimal transport, score matching, empirical Bayes estimation, and bridge matching frameworks.