Sutton's 15th PhD student publishes thesis on lifelong plasticity in neural networks
RichardSSutton · x · 2026-10-04
Reinforcement learning pioneer Richard Sutton announced that the PhD thesis of his 15th doctoral student, Shibhansh Dohare, is now available: Learning Forever using Artificial Neural Networks.
The dissertation centers on plasticity — the ability of neural networks to keep learning new things:
- Systems interacting with the real world become outdated unless they continually learn from new, partially observable information
- Naturally intelligent systems all learn continually to keep up with change
- Applications like market/weather prediction, learning human preferences, and controlling factories or prosthetics require AI that learns forever
- The thesis provides direct and thorough empirical demonstrations of this capability
A notable release for anyone following continual learning and plasticity loss research.
More from Research
- LEV: Looped decision model lifts 2.6B accuracy 63.2% to 76.2% with adaptive compute — jiqizhixin · 2026-11-19
- Aurora2-400M-A20M: a sparse MoE pretrained from scratch in 36 hours on one RTX Pro 6000 — Tall_Abrocoma_3533 · 2026-11-19
- Researcher calls METR/Anthropic '2x speedup' RSI analysis statistically dubious — jessi_cata · 2026-11-19
- Linear Probes Could Reveal Which Persona a Base Model Is Playing, Researchers Suggest — cephaloform · 2026-11-19
- ETH's Anymal robot completes 10km autonomous alpine hike with VLM-guided ROAM system — ChongZzZhang · 2026-11-19
- OpenVax codebase grows from 78K to 565K lines in 15 months — iskander · 2026-11-19