flow matching
A method used in generative models to estimate the similarity of probability distributions through the concept of optimal transport, enabling better modeling of complex data distributions.
- Aligning Text to Image in Diffusion Models is Easier Than You Think
- Blockwise Flow Matching: Improving Flow Matching Models For Efficient High-Quality Generation
- BrainFlow: A Holistic Pathway of Dynamic Neural System on Manifold
- Composite Flow Matching for Reinforcement Learning with Shifted-Dynamics Data
- Counterfactual Identifiability via Dynamic Optimal Transport
- Failure Prediction at Runtime for Generative Robot Policies
- Flow Matching Neural Processes
- Flow Matching-Based Autonomous Driving Planning with Advanced Interactive Behavior Modeling
- High-Order Flow Matching: Unified Framework and Sharp Statistical Rates
- High-order Equivariant Flow Matching for Density Functional Theory Hamiltonian Prediction
- Joint Velocity-Growth Flow Matching for Single-Cell Dynamics Modeling
- Knowledge Insulating Vision-Language-Action Models: Train Fast, Run Fast, Generalize Better
- Learning non-equilibrium diffusions with Schrödinger bridges: from exactly solvable to simulation-free
- Mean Flows for One-step Generative Modeling
- Mean Flows for One-step Generative Modeling
- Multivariate Latent Recalibration for Conditional Normalizing Flows
- On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity
- On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity
- On the Relation between Rectified Flows and Optimal Transport
- Overcoming Challenges of Long-Horizon Prediction in Driving World Models
- Path Gradients after Flow Matching
- Scalable, Explainable and Provably Robust Anomaly Detection with One-Step Flow Matching
- Sequence Modeling with Spectral Mean Flows
- Shallow Flow Matching for Coarse-to-Fine Text-to-Speech Synthesis
- Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free Lunch
- Show-o2: Improved Native Unified Multimodal Models
- Solving Inverse Problems with FLAIR
- Transition Matching: Scalable and Flexible Generative Modeling
- Whitened Score Diffusion: A Structured Prior for Imaging Inverse Problems