New DFMU method enables data-frugal machine unlearning

Researchers have introduced Data-Frugal Machine Unlearning (DFMU), a novel method designed to efficiently remove data elements from trained machine learning models. Unlike existing approaches that often require extensive retraining, DFMU utilizes a single forward and backward pass to compute an importance score for model components. This technique, based on knowledge-preserving pruning, significantly reduces computational needs and data requirements. Experiments show DFMU achieves 40% higher retain-accuracy with only 13% of the data compared to state-of-the-art methods, while also processing forgetting tasks 88% faster. AI

IMPACT This method could significantly reduce the computational cost and time required for model unlearning, making it more accessible for applications needing to remove specific data.

RANK_REASON The cluster describes a new research paper detailing a novel method for machine unlearning.

Read on arXiv cs.LG →

• Data-Frugal Machine Unlearning
• DFMU
• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Litmaps
• machine unlearning
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →