Projects
Projects are optional assignments for students who would like to earn extra points and explore one of the course methods in more depth. Students may work individually or in teams of up to five people.
Each project focuses on reproducing an existing method, running the available code, and evaluating the results with standard metrics on either an EOT or OT benchmark.
| No. | Project | Paper | Code | Status |
|---|---|---|---|---|
| 1 | ICNN-OT | Optimal Transport Mapping via Input Convex Neural Networks, ICML 2020 | OT-ICNN | Available |
| 2 | Diffusion Schrödinger Bridge (DSB) | Diffusion Schrödinger Bridge with Applications to Score-Based Generative Modeling, NeurIPS 2021 | diffusion_schrodinger_bridge | Available |
| 3 | Likelihood training of Schrödinger Bridge (FBSDE) | Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory, ICLR 2022 | SB-FBSDE | Available |
| 4 | Optimal Flow Matching (OFM) | Optimal Flow Matching: Learning Straight Trajectories in Just One Step, NeurIPS 2024 | Optimal-Flow-Matching | Vladislav Minashkin, Vadim Kasiuk |
| 5 | Simulation-free score and flow matching | Simulation-Free Schrödinger Bridges via Score and Flow Matching, AISTATS 2024 | conditional-flow-matching | Available |
Benchmarks
- Deterministic maps / flow: Wasserstein-2 Benchmark
- Entropic / Schrödinger-bridge-style: Entropic OT Benchmark