SoftServe's updates build on quadratic matrix equations, extending the authors' Batch-and-Match BBVI work
dianarycai · x · 2026-10-02
The author adds context to the SoftServe paper: its updates are based on solutions to quadratic matrix equations, extending the team's earlier ICML 2024 work Batch-and-Match (BaM), which showed that score matching in a Gaussian family reduces to solving a quadratic matrix equation.
BaM is a black-box variational inference method using a score-based divergence, with a closed-form proximal update for Gaussian families, exponential convergence in the infinite-batch limit, and typically fewer gradient evaluations than ELBO-based BBVI. This post is supplementary context for the same SoftServe work.
Related event: SoftServe: Quasi-Newton Optimization Scales to Massive Neural Networks(3 posts)→
More from Research
- AI 'speech clock' predicts how fast you're ageing from your voice, featured in Nature — AnnaCiaunica · 2026-10-02
- ScholarCatalyst: a new benchmark testing whether AI can pick research problems like humans — PangWeiKoh · 2026-10-02
- Clarification as Supervision lands NeurIPS Oral: denser training signal via model interaction — iatitov · 2026-10-02
- Ego-Exo4D-HM: SMPL-H reconstructions for 523 hours of egocentric-exocentric video, open-sourced — geopavlakos · 2026-10-02
- Cohere Labs Talk: Making AI Math Reasoning Machine-Checkable with Lean — Cohere_Labs · 2026-10-02
- New paper dissects only task-relevant network parts, making mechanisms inspectable and editable at far lower cost — leedsharkey · 2026-10-02