Skip to content

Search problems, genes, families, sources

Find a problem Families All problems Notation & genes Data & method About

The Scoreboard Is Corrupted

Machine Learning Benchmark Datasets Accumulate Errors That Distort Research Progress

digital-ml-benchmark-dataset-integrity-erosion

View full brief →

A living collection from the Altringer Eagle Lab at Dartmouth’s Thayer School of Engineering. Problems come from expert sources — research agendas, standards bodies, agencies — and every source is fetched and checked before a problem enters; independent readers re-tag samples blind so the genes can be trusted.

How verification works Data & method Bring us a problem Teach with the collection