Imagen de Portada

Two Birds, One Projection: Harmonizing Safety and Utility in LVLMs via Inference-time Feature Projection

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Han, Yewon, Seol, Yumin, Kong, EunGyung, Jo, Minsoo, Kim, Taesup
Formato:	Preprint
Publicado:	2026
Materias:	Computer Vision and Pattern Recognition Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Angular Gradient Sign Method: Uncovering Vulnerabilities in Hyperbolic Networks
por: Jo, Minsoo, et al.
Publicado: (2025)

Attention-space Contrastive Guidance for Efficient Hallucination Mitigation in LVLMs
por: Jo, Yujin, et al.
Publicado: (2026)

Generalized and Personalized Federated Learning with Black-Box Foundation Models via Orthogonal Transformations
por: Kong, Eun Gyung, et al.
Publicado: (2025)

Contrastive Residual Energy Test-time Adaptation
por: Han, Yewon, et al.
Publicado: (2025)

Retaining and Enhancing Pre-trained Knowledge in Vision-Language Models with Prompt Ensembling
por: Kim, Donggeun, et al.
Publicado: (2024)

Missing Modality Prediction for Unpaired Multimodal Learning via Joint Embedding of Unimodal Models
por: Kim, Donggeun, et al.
Publicado: (2024)

CL3DOR: Contrastive Learning for 3D Large Multimodal Models via Odds Ratio on High-Resolution Point Clouds
por: Kim, Keonwoo, et al.
Publicado: (2025)

Improving Visual Token Reduction via Rectifying Distortions for Efficient Multimodal LLM Inference
por: Cho, Hyeonwoo, et al.
Publicado: (2026)

Diffusion-Based Conditional Image Editing through Optimized Inference with Guidance
por: Lee, Hyunsoo, et al.
Publicado: (2024)

Semantic Anchoring for Robust Personalization in Text-to-Image Diffusion Models
por: Yang, Seoyun, et al.
Publicado: (2025)

Preserve and Personalize: Personalized Text-to-Image Diffusion Models without Distributional Drift
por: Kim, Gihoon, et al.
Publicado: (2025)

Video-SafetyBench: A Benchmark for Safety Evaluation of Video LVLMs
por: Liu, Xuannan, et al.
Publicado: (2025)

Diffusion-Based Image-to-Image Translation by Noise Correction via Prompt Interpolation
por: Lee, Junsung, et al.
Publicado: (2024)

PiCa: Parameter-Efficient Fine-Tuning with Column Space Projection
por: Hwang, Junseo, et al.
Publicado: (2025)

Memory-Free Continual Learning with Null Space Adaptation for Zero-Shot Vision-Language Models
por: Jo, Yujin, et al.
Publicado: (2025)

Patch-Level Kernel Alignment for Dense Self-Supervised Learning
por: Yeo, Juan, et al.
Publicado: (2025)

MPA-DNN: Projection-Aware Unsupervised Learning for Multi-period DC-OPF
por: Kim, Yeomoon, et al.
Publicado: (2025)

Projecting Gaussian Ellipsoids While Avoiding Affine Projection Approximation
por: Qi, Han, et al.
Publicado: (2024)

Q-Align: Alleviating Attention Leakage in Zero-Shot Appearance Transfer via Query-Query Alignment
por: Kim, Namu, et al.
Publicado: (2025)

MAFA: Managing False Negatives for Vision-Language Pre-training
por: Byun, Jaeseok, et al.
Publicado: (2023)

Mitigating Object Hallucinations in LVLMs via Attention Imbalance Rectification
por: Sun, Han, et al.
Publicado: (2026)

Incremental Learning with Repetition via Pseudo-Feature Projection
por: Tscheschner, Benedikt, et al.
Publicado: (2025)

Sparse Imagination for Efficient Visual World Model Planning
por: Chun, Junha, et al.
Publicado: (2025)

Overcoming Data Inequality across Domains with Semi-Supervised Domain Generalization
por: Park, Jinha, et al.
Publicado: (2024)

PosterLlama: Bridging Design Ability of Langauge Model to Contents-Aware Layout Generation
por: Seol, Jaejung, et al.
Publicado: (2024)

Steering LVLMs via Sparse Autoencoder for Hallucination Mitigation
por: Hua, Zhenglin, et al.
Publicado: (2025)

DoMIX: An Efficient Framework for Exploiting Domain Knowledge in Fine-Tuning
por: Kim, Dohoon, et al.
Publicado: (2025)

V-Attack: Targeting Disentangled Value Features for Controllable Adversarial Attacks on LVLMs
por: Nie, Sen, et al.
Publicado: (2025)

Feature Projection Learning for Better Vision-Language Reasoning
por: Zhang, Yi, et al.
Publicado: (2026)

Catch-Up Mix: Catch-Up Class for Struggling Filters in CNN
por: Kang, Minsoo, et al.
Publicado: (2024)

TARAC: Mitigating Hallucination in LVLMs via Temporal Attention Real-time Accumulative Connection
por: Jiang, Lei, et al.
Publicado: (2025)

Object-Centric World Model for Language-Guided Manipulation
por: Jeong, Youngjoon, et al.
Publicado: (2025)

DistilDIRE: A Small, Fast, Cheap and Lightweight Diffusion Synthesized Deepfake Detection
por: Lim, Yewon, et al.
Publicado: (2024)

Seeing Clearly by Layer Two: Enhancing Attention Heads to Alleviate Hallucination in LVLMs
por: Zhang, Xiaofeng, et al.
Publicado: (2024)

HarmoniCa: Harmonizing Training and Inference for Better Feature Caching in Diffusion Transformer Acceleration
por: Huang, Yushi, et al.
Publicado: (2024)

EDM: Equirectangular Projection-Oriented Dense Kernelized Feature Matching
por: Jung, Dongki, et al.
Publicado: (2025)

TLDR: Text Based Last-layer Retraining for Debiasing Image Classifiers
por: Park, Juhyeon, et al.
Publicado: (2023)

DEAL: Decoupled Classifier with Adaptive Linear Modulation for Group Robust Early Diagnosis of MCI to AD Conversion
por: Lee, Donggyu, et al.
Publicado: (2024)

TFLOP: Table Structure Recognition Framework with Layout Pointer Mechanism
por: Khang, Minsoo, et al.
Publicado: (2025)

ATAS: Any-to-Any Self-Distillation for Enhanced Open-Vocabulary Dense Prediction
por: Yeo, Juan, et al.
Publicado: (2025)