Helmholtz-SDE closes the approximation gap in simulation-free latent SDEs
scott_linderman · x · 2026-10-13
A new NeurIPS 2026 paper from Scott Linderman's group (led by Henry Smith) introduces Helmholtz-SDE, a faster and more accurate simulation-free inference algorithm for latent SDEs.
Key points:
- Existing simulation-free VI methods parameterize the posterior via instantaneous marginals, but the authors show this restricts the approximate posterior to a subset of SDEs reachable by simulation-based methods, degrading inference and parameter learning
- Helmholtz-SDE closes this gap by optimizing over path laws compatible with a prescribed collection of marginals
- Largest gains under high posterior uncertainty; matches simulation-based VI at a fraction of the runtime
Paper and code are publicly available.
More from Research
- DUET paper co-evolves agent and LLM grader, hits 96.5% on Finance Agent benchmark — dair_ai · 2026-10-13
- Base Labs investigates why models forget facts they iteratively learn, with 13B talkie — baseten · 2026-10-13
- Robot deployment is becoming the operating layer for post-training physical AI policies — paigeinsf · 2026-10-13
- New J++ Lens interpretability method delivers 55% more faithful readouts than J-Lens — wesg52 · 2026-10-13
- Spatial biology slide costs stuck at $5-10k for five years, defying Moore's law — anshulkundaje · 2026-10-13
- Google Research trains user simulators with latent intrinsic feedback, cutting RL compute by over half — cocoweixu · 2026-10-13