Arc contest winner: our method is a denoised weighted mean, 2× simple baseline

kjaganatha · x · 2026-11-20

Responding to Stanford professor Anshul Kundaje's skepticism that the solution is a nearest-neighbor-style shortcut, the author of the Arc Institute cell-simulation contest winning entry says the core of the method is a denoised weighted mean that achieved roughly 2× the performance of a simple mean on the metrics.

Related event: Stanford professor challenges Arc virtual cell challenge winner: data privilege or methodological edge(18 posts)→

Original post →