Topic
Neural network architectures that learn from graph-structured data, capturing relationships between nodes for classification and prediction tasks.
SpaCEy, a graph neural network in Nature Communications, sorted breast cancer patients by survival from spatial protein maps without cell-type labels. Its split held within established subtypes, the one place it could add prognostic information clinics lack.
Reality
Tautomer assignment has been inference wherever protein crystallography cannot see hydrogen atoms. An NYU group borrowed a million experimental answers from small-molecule crystals and trained a network to make the call.
Reality
A Leibniz Institute pipeline learned from roughly 500 known bitter peptides, invented 161 more, and sent the most distinctive 31 to a blinded trained panel that agreed with the model on 25. The scoring was bitter or not bitter.
Reality
Cloudflare says its graph model flagged eight malicious scripts running on live storefronts. A retrospective check put seven of the eight outside VirusTotal entirely. URLScan flagged none of the eight as malicious.
Reality
LF Energy took in four projects on September 15th and filed all of them at its entry tier. AssetLife arrives with four operators running it, while EnerGNN's 200-millisecond result found less capacity than the solver it beat.
Reality
A Michigan-led ATLAS analysis claims world-best sensitivity to the Higgs boson interacting with itself, using machine learning on data already in hand, four years before the LHC upgrade.
Reality