Ten years in deep learning: back then CS231n opened with an apology for whether DL worked
_aidan_clark_ · x · 2026-09-25
Researcher Aidan Clark reflects on a decade in deep learning: ten years ago he took John Canny's Cal-ification of CS231n, a course that opened with an apology—enough people, including the professor he worked with, still doubted deep learning actually worked. He also wrote a 'terrible' but prescient paper on low-precision training. Looking back, many speakers and cited authors are now people he knows personally.
More from Companies & People
- Elon Musk reshares old post warning against complacency and entitlement — shaunmmaguire · 2026-09-25
- Economist Study: AI Boosts Homework Scores 18% but Exam Scores Fall 20% — robleclerc · 2026-09-25
- Meta's Muse agent looks a lot like OpenClaw, fueling wrapper-app claims — haydenfield · 2026-09-25
- NeurIPS desk-rejects 5/5/5-scored paper over one co-author's policy violation — AccBalanced · 2026-09-25
- Stanford launches CS329Z, a full course on engineering AI agents — stanfordnlp · 2026-09-25
- Ollama and Crusoe to debate closed vs open vs self-trained models in SF panel — ollama · 2026-09-25