Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Machine Unlearning on Convex cubic loss d=1 200,000 (test)

0.0017Sigma

Perturbed Newton

-0.023020.143840.31070.47756Jul 7, 2026
Updated 18d ago

Evaluation Results

MethodLinks
2026.07
0.001710.19839.4159
2026.07
0.001710.19839.4159
2026.07
0.0028.52376.7425
2026.07
0.0028.52376.7425
2026.07
0.00256.23774.3663
2026.07
0.00256.23774.3663
2026.07
0.00363.7352.1455
2026.07
0.00363.7352.1455
2026.07
0.00542.33221.0504
2026.07
0.00542.33221.0504
2026.07
0.00811.47270.4827
2026.07
0.00811.47270.4827
2026.07
0.0370.29020.0141
2026.07
0.0370.29020.0141
2026.07
0.619700
2026.07
0.619700