20 results for “Forgetting”
CS papers onlyHybrid search: Keyword + semantic, ranked by combined score.ⓘ
Want pure semantic search? Try claim verification →
Dayong Ye, Tainqing Zhu, Congcong Zhu, Feng He +4 more
The paper proposes a comprehensive framework for LLM-based agent unlearning, enabling agents to selectively forget specific knowledge (states, trajectories, or environments) while maintaining performa…
This paper presents an interactive installation that visualizes the subjective experience of forgetting through a neural network's degeneration.
PURGE is a novel machine unlearning algorithm that leverages the duality between continual learning and unlearning to achieve high data retention while making the unlearned model indistinguishable fro…
Dongjun Kim, Adrian de Wynter, Huancheng Chen, Heasung Kim +1 more
The paper introduces FoLoRA, a novel optimization framework that uses a generalized Rayleigh quotient to achieve a superior balance between adapting foundation models to specific tasks and preserving…
This paper investigates the geometric structure of recoverability in continual learning and challenges the Recoverability Diffusion hypothesis.
This paper introduces the first complete pipeline for federated unlearning, proposing an efficient unlearning approach and a novel visualization framework (Skyeye) to evaluate a model's forgetting cap…
This paper introduces Anchored Weight Decay (AWD), a regularization technique that effectively prevents prior-task forgetting during LLM fine-tuning with Evolution Strategies (ES), positioning ES as a…
Suryash Yagnik, Shubham Gaur, Saksham Thakur, Vinija Jain +2 more
The paper introduces 5WBENCH, a new benchmark for causal unlearning, and proposes MAAT, a novel three-phase framework that achieves high forgetting and high retention specifically on complex 'Why'-typ…
This paper investigates how Batch Normalization affects inference in edge intelligence systems and proposes a method to audit state integrity.
The paper introduces AMNESIA, the first large-scale, open-source benchmark for medical unlearning, demonstrating that current unlearning methods struggle to separate individual patient data from share…
Xukun Luan, Zhongxiang Lei, Chen Gong, Shaowei Li +2 more
The paper proposes a method called ForgetMimic to eliminate specific motions from learned policies in humanoid control while preserving the effectiveness of other motions.
Mengying Zhang, Derui Wang, Ruoxi Sun, Xiaoyu Xia +2 more
This paper provides the first integrated analysis of model dememorization, unifying unlearnability and unlearning methods, and offering theoretical guarantees on dememorization depth.
This paper proposes a system called \sys{} that governs an agent's experience-memory lifecycle by scoring each entry as value minus harm per byte, reducing memory, uplink, and injection success while…
This paper tests Memoir, a neural network model with per-sample fast memory, shared slow parameters, variable-depth latent recurrence, and a future-latent energy objective, and compares it to a read-o…
Lei, Chen, Guilin Zhang, Kai Zhao +7 more
This paper studies the impact of memory design choices on the trade-off between personalization utility, extraction risk, and deletion fidelity in long-lived foundation-model agents.
The paper demonstrates that self-reflective agents can systematically confabulate incorrect memories, leading them to fail tasks even when the environment resets, and proposes a metric and mitigation…
This paper proposes a sparse cache with a novel allocation rule based on DP-means clustering for deep recurrent models, achieving full associative recall with only distinct items, outperforming other…