Seonglae ChoAn obvious alternative to unlearning is to not learn at all (RAG)
- Exact unlearning (distributionally identical) Differential privacy
- Empirical unlearning (unlearning via fine-tuning)
Machine Unlearning Notion
Machine Unlearning Methods
Open problems
arxiv.org
https://arxiv.org/pdf/2501.04952
Unlearning offers an incomplete solution for impermissible knowledge removal in LLMs with strong In-context learning capability
arxiv.org
https://arxiv.org/pdf/2407.00106
arxiv.org
https://arxiv.org/pdf/2501.04952
LLMs can approximate unlearning when prompted with instructions and demonstrations. (Trivial Prompting)
arxiv.org
https://arxiv.org/pdf/2310.07579