Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning-to-Optimize via Deep Unfolded Flows
Augustinos D Saravanos, Oswin So, H M Sabbir Ahmad, Chuchu Fan
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 FlowOptimizer instead of best baseline (MS-Adam) on power grid dispatch | FlowOptimizer · Power Grid (n=50) | Best(f(X_K))/f(mu_0): 0.05 -> 1.2e-3 | Improved |
| use FlowOptimizer instead of PSO on robotic arm IK | FlowOptimizer · Robotic Arm (n=30) | Best(f(X_K))/f(mu_0): 0.011 -> 0.007 | Improved |
| use FlowOptimizer instead of L2O-GD on supply chain | FlowOptimizer · Supply Chain (n=80) | Best(f(X_K))/f(mu_0): 0.027 -> 0.013 | Improved |
| population-level flow update instead of learned single-point GD steps (L2O-GD) | FlowOptimizer · Ackley/Rastrigin/Levy 20D | best f error | Improved |
| dimension-agnostic variant trained on 20D, deployed on 200D | FlowOptimizer · Ackley 200D, Levy 200D | best f error | Improved |
| add optional Gaussian base sampler (ablation) | FlowOptimizer · benchmarks | stability | Mixed |
| use CEM and random search on high-dimensional multimodal problems | CEM, RS · Ackley/Rastrigin/Levy 20D | best f error | Hurt |
Closest papers by meaning
- Understanding Optimization in Deep Learning with Central Flows (ICLR 2025 · similarity 0.88)
- Solver-Free Decision-Focused Learning for Linear Optimization Problems (NeurIPS 2025 · similarity 0.87)
- ParetoFlow: Guided Flows in Multi-Objective Optimization (ICLR 2025 · similarity 0.86)
- Investigation into Training Dynamics of Learned Optimizers (Student Abstract) (AAAI 2024 · similarity 0.85)
- Flow Density Control: Generative Optimization Beyond Entropy-Regularized Fine-Tuning (NeurIPS 2025 · similarity 0.85)
- Optimal Flow Matching: Learning Straight Trajectories in Just One Step (NeurIPS 2024 · similarity 0.85)
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.
