RL lecturer shares lecture slides: supervised learning approximates, RL discovers functions
KyleMorgenstein · x · 2026-11-18
Continuing his "intelligence is RL" argument, UT lecturer Kyle Morgenstein writes: "this isn't some big secret — these are the first slides of my RL lecture notes that I teach at UT."
Key points (with slide screenshots):
- Supervised learning ⇔ function approximation: dataset + hypothesis class + minimizing prediction error
- Reinforcement learning ⇔ function discovery: trajectories + policy + maximizing expected discounted reward
- Function approximation works if you already know what you want; intelligence is about function discovery — if you can write down a (proxy) objective, you can optimize it
The post grounds the whole thread in his course material, formalized via the Agent-Environment loop diagram.
More from Research
- Eval methodology: statistical tests are valid, but conclusions must span prompt, judge, and harness variance — srchvrs · 2026-11-18
- Supervised vs self-supervised is a false dichotomy: RL is where it ends up — KyleMorgenstein · 2026-11-18
- "Supervised vs self-supervised" is a false dichotomy, argues UT RL lecturer — KyleMorgenstein · 2026-11-18
- ELLIS Sites NeurIPS Fest 2026 returns across Europe with 18 accepted NeurIPS papers — DocXavi · 2026-11-18
- Decompiled Instagram APK reveals on-device models and probing for 23 installed AI apps — ChenhaoTan · 2026-11-18
- Running a 180B MoE Model on a 16GB MacBook: Experts Offloaded to SSD, 14× Faster Than llama.cpp — XiongChenyan · 2026-11-18