langevin dynamics
This is a technique used in machine learning for sampling from distributions using stochastic gradients, which provides a way to explore the parameter space while incorporating noise, often aimed at achieving a balanced trade-off between exploration and exploitation.
- Continuous-time Riemannian SGD and SVRG Flows on Wasserstein Probabilistic Space
- Diffusion Generative Modeling on Lie Group Representations
- Fast Non-Log-Concave Sampling under Nonconvex Equality and Inequality Constraints with Landing
- Fractional Langevin Dynamics for Combinatorial Optimization via Polynomial-Time Escape
- Mitigating Instability in High Residual Adaptive Sampling for PINNs via Langevin Dynamics
- ORIGEN: Zero-Shot 3D Orientation Grounding in Text-to-Image Generation
- PID-controlled Langevin Dynamics for Faster Sampling on Generative Models
- Posterior Sampling by Combining Diffusion Models with Annealed Langevin Dynamics
- Preconditioned Langevin Dynamics with Score-based Generative Models for Infinite-Dimensional Linear Bayesian Inverse Problems
- Temperature is All You Need for Generalization in Langevin Dynamics and other Markov Processes