The paper proposes Orthogonal Concept Erasure (OCE), a novel multiplicative parameter update method that achieves precise concept erasure in diffusion models while independently preserving overall gen…

The paper introduces TICoE, a text-image collaborative framework that achieves precise and faithful concept removal from text-to-image generative models, surpassing existing methods in both precision…

CoreUnlearn introduces a novel framework that disentangles and removes undesirable concepts from text-guided diffusion models by targeting specific, erasure-critical components of the concept embeddin…

The paper proposes AHV-D&S, a novel training-free inference-time safeguard that detects and suppresses risky content in Diffusion Transformers (DiTs) by quantifying token sensitivity across attention…

The paper introduces GEM, an effective concept erasure framework for Rectified Flow Transformers, by unifying trajectory-based unlearning with classic teacher-guided flow matching.

The paper proposes Hydra, a framework to stabilize and control the injection of multiple, conflicting backdoor triggers into text-to-image diffusion models, ensuring high attack reliability while main…

The paper proposes a disentangled representation framework to significantly improve few-shot layout-to-image generation by separating semantic identity from local visual details, thereby mitigating re…

This paper systematically investigates unlearnable examples (UEs) across diverse training paradigms, finding that existing UEs fail under pretraining-finetuning (PF) settings, and proposes Shallow Sem…

Equilibrated Diffusion introduces a frequency-aware approach to image customization, disentangling style and subject content embeddings to achieve superior subject fidelity and text adherence.

The paper proposes Locality-Aware Redundancy Pruning (LoRP), a training-free method that prunes LLM layers by exploiting localized inter-layer redundancy, leading to improved efficiency while maintain…

AnchorSteer introduces a framework that achieves high-fidelity, structure-preserving music editing by decoupling semantic concept injection from structural constraints.

The paper proposes a unified, constrained optimization framework using KL divergence and likelihood constraints to achieve effective and principled unlearning in diffusion models.

The paper deconstructs latent visual reasoning tokens into components and finds that the performance gains are primarily due to boundary markers and attention patterns, not the tokens' ability to enco…

The paper demonstrates that content suppression techniques used in language models only mask prohibited content at the output level, failing to eliminate the underlying concepts from the model's inter…

This paper demonstrates that Concept Bottleneck Models (CBMs), despite their interpretability, are highly vulnerable to targeted adversarial attacks that manipulate semantic concepts, and proposes SPE…

肖代替了视觉令牌的永久删除，通过可恢复的路由来改进视觉语言模型的性能

The paper introduces SARDI, a novel, training-free framework that uses low-confidence 'lookahead' tokens generated during the denoising process of discrete diffusion language models to dynamically gui…

The paper proposes a novel global sketch-based watermarking technique for diffusion language models that controls the entire sequence's statistics, offering an order-agnostic and context-independent a…

LoRA-Key introduces a user-centric watermarking framework that attaches a recoverable ownership key to LoRA modules via a standalone Watermark LoRA, providing lightweight, plug-and-play copyright prot…