Apple Proposes Low-Cost Machine Unlearning: Skip Retraining for Low-Influence Data
Apple ML Research · rss · 2026-08-13
Apple ML Research published a new paper tackling the challenge of machine unlearning. As data privacy concerns grow, safely removing specific data points from trained models has become critical.
The authors challenge the conventional approach of treating all forget-set data equally. By analyzing influence functions across language and vision tasks, the research identifies training subsets with negligible impact on model outputs. The paper suggests that these "low-influence points" can be skipped during retraining, drastically reducing the computational costs associated with machine unlearning.
More from Research
- Square lattice connective constant remains unsolved — ripe for AI, argues mathematician — tak3sh8 · 2026-10-16
- Quiz: Double the Residual Stream from 4096 to 8192 — How Does Param Count Change? — Sauers_ · 2026-10-16
- Researcher credits GPT for multiple accepted papers, incl. disproving Wegner's 1965 conjecture — ctjlewis · 2026-10-16
- AutoAgent0: VLM-powered agentic runtime lifts long-tail driving success by 28.3pp — yining_hong · 2026-10-16
- Shuffled labels hit AUC 0.59: selection bias, not signal, drives prosody benchmarks — maier_ak · 2026-10-16
- CODESTRUCT reframes codebases as AST action spaces to fix brittle LLM code agents — burkov · 2026-10-16