{"slug": "unact-gradient-free-unlearning-via-targeted-activation-intervention", "title": "UnAct: Gradient-Free Unlearning via Targeted Activation Intervention", "summary": "Researchers introduced UnAct, a gradient-free machine unlearning method that removes the influence of designated training data by intervening on targeted activations rather than retraining or backpropagating. UnAct is positioned against retrain-free baselines Selective Synaptic Dampening (SSD) and its label-free variant LFSSD, which the authors note still require backpropagation and parameter updates.", "body_md": "Machine unlearning seeks to remove the influence of designated training data from a trained model without retraining from scratch. Retrain-free methods such as Selective Synaptic Dampening (SSD) and its label-free variant LFSSD avoid full retraining but still require backpropagation and parameter im", "url": "https://wpnews.pro/news/unact-gradient-free-unlearning-via-targeted-activation-intervention", "canonical_source": "https://aiflash.com/news/132440/", "published_at": "2026-10-07 01:30:19+00:00", "updated_at": "2026-10-07 01:48:02.731791+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "artificial-intelligence"], "entities": ["UnAct", "Selective Synaptic Dampening", "LFSSD"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/unact-gradient-free-unlearning-via-targeted-activation-intervention", "markdown": "https://wpnews.pro/news/unact-gradient-free-unlearning-via-targeted-activation-intervention.md", "text": "https://wpnews.pro/news/unact-gradient-free-unlearning-via-targeted-activation-intervention.txt", "jsonld": "https://wpnews.pro/news/unact-gradient-free-unlearning-via-targeted-activation-intervention.jsonld"}}