Researcher: forcing convex optimization onto ML solutions is like sanding blocks to fit shape-sorter holes
caglar_ee · x · 2026-11-09
An EE researcher recalls trying for a long time to make sense of certain ML solutions through a convex optimization lens — alternating minimization? some approximation? — but nothing fit. He likens it to a child playing with a shape-sorting toy, eventually sanding down the corners of wooden pieces to force them into the holes: a vivid self-deprecating take on shoehorning classical math frameworks onto modern ML.
Related event: Researcher: Classic ML Closed-Form Solutions Don't Rely on Convexity(2 posts)→
More from Fun
- Solo dev's infinitely evolving Claude-built Backrooms game hits 35K players in 36 hours — minchoi · 2026-11-09
- Musk touts Grok after user says it helped expose a PI tailing him via Raspberry Pi tool — elonmusk · 2026-11-09
- RentAHuman lets AI agents hire real people for physical-world tasks via MCP API — conitzer · 2026-11-09
- LukeW mocks AI startup pitch escalation: 'oh crap, we're SaaS' — LukeW · 2026-11-09
- Ex-Apple engineer: Siri hired a native Italian speaker for QA back in 2014 — jdluk87 · 2026-11-09
- Impatient waiting for the electrician, ML engineer grabs a few B200s from Lambda — TheZachMueller · 2026-11-09