Are deep ResNets provably better than linear predictors?

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.