The Approximately Correct Machine Intelligence (ACMI) Lab at @mldcmu at @SCSatCMU. Growing the ML sandbox to address more of the real world. PI @zacharylipton
May 8, 2021 • 5 tweets • 2 min read
New @ #ICML2021: When a trained model fits clean (training) data well but randomly labeled (training) data (added in) poorly, its generalization (to the population) is guaranteed!
by ACMI PhD @saurabh_garg67, Siva B, @zicokolter, & @zacharylipton
This result makes deep connections between label noise, early learning, and generalization. Key takeaways: 1) the early learning phenomenon can be leveraged to produce post-hoc generalization certificates; 2) can be leveraged by adding unlabeled training data (randomly labeled)
