Research map / Optimization / Optimizers
Learning of Population Dynamics: Inverse Optimization Meets JKO Scheme
Mikhail Persiianov, Jiawei Chen, Petr Mokrov, Alexander Tyurin, Evgeny Burnaev, Alexander Korotin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use iJKOnet_{t,V} instead of JKOnet*_{t,V} | time-varying potential · EB 5D t1 | W2 distance: 4.414 -> 0.983 | Improved |
| use iJKOnet_V instead of JKOnet*_V | static potential · EB 5D t3 | W2 distance: 2.529 -> 1.147 | Improved |
| use iJKOnet_V instead of JKOnet*_V | static potential · EB 100D w/o LO | MMD: 0.229 -> 0.085 | Improved |
| use iJKOnet_V instead of DMSB | JKO-based model · EB 100D w/o LO | MMD: 0.032 -> 0.085 | Hurt |
| add interaction and internal energies to potential (full iJKOnet) (ablation) | iJKOnet · EB 5D t1 | W2 distance: 1.082 -> 3.577 | Hurt |
| add internal energy to potential (ablation) | iJKOnet · EB 5D t1 | W2 distance: 1.082 -> 1.065 | No clear effect |
| use iJKOnet_{t,V} instead of iJKOnet_V on LO-t1 (ablation) | iJKOnet · EB 100D LO-t1 | MMD: 0.137 -> 0.848 | Hurt |
| switch from paired to unpaired samples (ablation) | iJKOnet_V · 2D potentials (Watershed) | B-dW2-UVP ratio: ratio 18135 | Hurt |
| use iJKOnet_W instead of JKOnet*_W for interaction energy | interaction energy · Flowers unpaired 10K | L2-UVP: 0.0069 -> 0.0009 | Improved |
Closest papers by meaning
- Learning diffusion at lightspeed (NeurIPS 2024 · similarity 0.87)
- Parametric model reduction of mean-field and stochastic systems via higher-order action matching (NeurIPS 2024 · similarity 0.87)
- Wasserstein residuals: Learning Gradient Flows from Population Dynamics (NeurIPS 2026 · similarity 0.86)
- Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics (ICLR 2025 · similarity 0.84)
- Diffusion Models are Evolutionary Algorithms (ICLR 2025 · similarity 0.84)
- Learning to Sample From Diffusion Models via Inverse Reinforcement Learning (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
