Unverified50% confidenceFactExact time
The core question posed by this study is: when a gradient-based unlearner differs from a model retrained only on the retained data, what structure does this output-space gap have.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedExisting gradient-based unlearning methods account only partially for the geometric relationships characterized by the leverage matrices, causing them to deviate significantly from the retrained model in output-space behavior.79% similarUnverifiedThis study reveals a fundamental disconnect between loss and behavior in gradient-based machine unlearning and provides a unified output-space analysis framework.78% similarUnverifiedThe core mechanism of experience replay is to maintain a memory buffer of limited capacity that stores some samples from historical tasks, and when training on a new task, mix them with the current data to form mini-batches, so that the loss function always contains gradient signals from the old distribution.75% similarUnverifiedFine-tuning and persistent memory solve different problems: fine-tuning changes model behavior through training samples, while persistent memory lets the application store information externally and retrieve it when needed.75% similarUnverifiedThis gap stems from a bias-variance trade-off in spectral space, for which the research team provides a closed-form characterization; learning β from data achieves automatic spectral regularization.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1113841API
curl https://kongchang.com/api/v1/knowledge/claims/1113841MCP
get_claim(id=1113841)