Entity · technique

ASAM

techniqueactiveasam-83ab6127·1 events·first seen May 25, 2026

Aliases: ASAM

Co-occurring entities

Multimodal Large Language Models Latent Adversarial Robustification (LAR)knowledge editing Rank-Constrained Subspace Learning (RCSL)

More like this (12)

AMIA ActiveSAM FunASR AMRS AMEL MedSAM AASIST DAAM ASAP++SAMPA SEAME UltraSAM

Recent events (1)

5arXiv · cs.AI·May 25, 2026·source ↗

Adversarial Subspace Alignment for Robust Multimodal Knowledge Editing in MLLMs

This paper addresses the generalization gap in multimodal large language model (MLLM) knowledge editing, where edits fail to propagate across semantically equivalent visual and linguistic variations. The authors introduce Latent Adversarial Robustification (LAR), which generates adversarial but semantically coherent variants in joint latent space, and Rank-Constrained Subspace Learning (RCSL), which enforces low-rank alignment of adversarial representations at the edit layer. Together these form the ASAM framework, which formalizes robustness via knowledge units grouping semantically equivalent multimodal inputs. Empirical analysis demonstrates improved generality without sacrificing reliability or locality.

Alignment and RLHF Multimodal Progress Multimodal Large Language Models Latent Adversarial Robustification (LAR)knowledge editing +2 more