Contestant fires back: denoising and weighting are a real methodological contribution

kjaganatha · x · 2026-11-20

Responding to Anshul Kundaje's critique, contestant kjaganatha argues their method achieves 2× the mean baseline performance with dataset and evaluation held constant, and that the denoising and weighting steps enabling this are a genuine methodological contribution.

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

Original post →