AI discovers an improved e-means in 10 minutes after counterexamples seemed to doom it
IgorCarron · x · 2026-11-19
A near real-time algorithm evolution: k-means++ has been the clustering mainstay for a decade; e-means came out last week; yesterday two counterexamples seemed to show e-means wasn't actually better; and today (per the quoted post) researchers used 6.2 Astra to discover an improved e-means that may get around both counterexamples within 10 minutes, with a report being written up. The author highlights the speed of discovery itself — AI is compressing academic iteration cycles to days or even minutes.
More from Research
- Universal Data-Efficient Pretraining Claims 56x Gains on Molecules, 24x on DNA and Crystals — madhavsinghal_ · 2026-11-19
- Fine-tuned World Action Model Dream-1 Jumps 159%, Beats Frontier LLMs on Robot Tasks — paigeinsf · 2026-11-19
- Fine-tuned Dream-1 world action model jumps 159% on real-robot benchmark at 1/100th the cost — paigeinsf · 2026-11-19
- AI makes 'living papers' nearly free to build, but their durability is doubtful — _akpiper · 2026-11-19
- HiT image generator hits 1.48 FID learning hierarchy from pixels, fast reproduced — niloofar_mire · 2026-11-19
- DaisyRose-27B: RL on 144 H200s for 18 days lifts Terminal-Bench from 5.6% to 12.0% — ying11231 · 2026-11-19