Functional Gradient Descent with Adaptive Representations accepted at NeurIPS
CatAstro_Piyush · x · 2026-09-27
The authors announce that "Functional Gradient Descent with Adaptive Representations" has been accepted at NeurIPS. Functional GD algorithms generally outperform neural nets but are hard to implement accurately; this work proposes adaptive representations to fix that, making functional optimization practical.
More from Research
- Rufus-Air paper: ordering post-training by reward reliability builds competitive open LLM — burny_tech · 2026-09-27
- Researchers teach LLMs to find interesting theorems, boosting discovery 4.3x — burny_tech · 2026-09-27
- DeepMind's Economic Policy for AGI framework evaluates 11 interventions — CurieuxExplorer · 2026-09-27
- Contrastive World Models Learns World Models in Latent Space Without Pixel Prediction — burny_tech · 2026-09-27
- Rethinking on-policy distillation: researchers propose OLIVE, letting students learn from teacher continuations — May_F1_ · 2026-09-27
- CoRL 2026 workshop on continually self-improving robots opens call for papers, due Sep 28 — PeterStone_TX · 2026-09-27