Vector Institute Shares 73 Papers Accepted to ICML 2026

VectorInst · x · 2026-07-10

The Vector Institute shared 73 papers its research teams will present at the upcoming ICML 2026, noting that machine learning is significantly accelerating the pace of scientific discovery.

Among them, they highlighted dnaHNet, a genomics foundation model developed by Bo Wang's team. This model breaks the limitations of fixed vocabularies by enabling dynamic DNA segmentation. Experiments show that it outperforms current leading models in processing efficiency and accuracy when predicting the effects of protein mutations and gene essentiality.

Original post →

More from Research

Research channel →