Imagen de Portada

Register and [CLS] tokens yield a decoupling of local and global features in large ViTs

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Lappe, Alexander, Giese, Martin A.
Formato:	Preprint
Publicado:	2025
Materias:	Computer Vision and Pattern Recognition Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Another BRIXEL in the Wall: Towards Cheaper Dense Features
por: Lappe, Alexander, et al.
Publicado: (2025)

STRAP-ViT: Segregated Tokens with Randomized -- Transformations for Defense against Adversarial Patches in ViTs
por: Chattopadhyay, Nandish, et al.
Publicado: (2026)

Intriguing Frequency Interpretation of Adversarial Robustness for CNNs and ViTs
por: Chen, Lu, et al.
Publicado: (2025)

Token Cropr: Faster ViTs for Quite a Few Tasks
por: Bergner, Benjamin, et al.
Publicado: (2024)

Adapting Pretrained ViTs with Convolution Injector for Visuo-Motor Control
por: Hwang, Dongyoon, et al.
Publicado: (2024)

Octic Vision Transformers: Quicker ViTs Through Equivariance
por: Nordström, David, et al.
Publicado: (2025)

Training-Free Acceleration of ViTs with Delayed Spatial Merging
por: Heo, Jung Hwan, et al.
Publicado: (2023)

Causality $\neq$ Decodability, and Vice Versa: Lessons from Interpreting Counting ViTs
por: Huang, Lianghuan, et al.
Publicado: (2025)

Exploring the Synergies of Hybrid CNNs and ViTs Architectures for Computer Vision: A survey
por: Yunusa, Haruna, et al.
Publicado: (2024)

ConcatPlexer: Additional Dim1 Batching for Faster ViTs
por: Han, Donghoon, et al.
Publicado: (2023)

DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs
por: Kim, Donghyun, et al.
Publicado: (2024)

Communication Efficient Split Learning of ViTs with Attention-based Double Compression
por: Alvetreti, Federico, et al.
Publicado: (2025)

TAP-ViTs: Task-Adaptive Pruning for On-Device Deployment of Vision Transformers
por: Wang, Zhibo, et al.
Publicado: (2026)

Decomposing and Interpreting Image Representations via Text in ViTs Beyond CLIP
por: Balasubramanian, Sriram, et al.
Publicado: (2024)

Concept-Guided Fine-Tuning: Steering ViTs away from Spurious Correlations to Improve Robustness
por: Elisha, Yehonatan, et al.
Publicado: (2026)

ViTCAE: ViT-based Class-conditioned Autoencoder
por: Jebraeeli, Vahid, et al.
Publicado: (2025)

UniRefiner: Teaching Pre-trained ViTs to Self-Dispose Dross via Contrastive Register
por: Qiu, Congpei, et al.
Publicado: (2026)

Parallel Backpropagation for Shared-Feature Visualization
por: Lappe, Alexander, et al.
Publicado: (2024)

Purrturbed but Stable: Human-Cat Invariant Representations Across CNNs, ViTs and Self-Supervised ViTs
por: Shah, Arya, et al.
Publicado: (2025)

I&S-ViT: An Inclusive & Stable Method for Pushing the Limit of Post-Training ViTs Quantization
por: Zhong, Yunshan, et al.
Publicado: (2023)

How to train your ViT for OOD Detection
por: Mueller, Maximilian, et al.
Publicado: (2024)

HydraViT: Stacking Heads for a Scalable ViT
por: Haberer, Janek, et al.
Publicado: (2024)

U-REPA: Aligning Diffusion U-Nets to ViTs
por: Tian, Yuchuan, et al.
Publicado: (2025)

Elastic ViTs from Pretrained Models without Retraining
por: Simoncini, Walter, et al.
Publicado: (2025)

Harnessing the Computation Redundancy in ViTs to Boost Adversarial Transferability
por: Liu, Jiani, et al.
Publicado: (2025)

Pretrained ViTs Yield Versatile Representations For Medical Images
por: Matsoukas, Christos, et al.
Publicado: (2023)

Colinearity Decay: Training Quantization-Friendly ViTs with Outlier Decay
por: Tong, Jin, et al.
Publicado: (2026)

Exploiting Lightweight Hierarchical ViT and Dynamic Framework for Efficient Visual Tracking
por: Kang, Ben, et al.
Publicado: (2025)

CubistMerge: Spatial-Preserving Token Merging For Diverse ViT Backbones
por: Gong, Wenyi, et al.
Publicado: (2025)

ProtoS-ViT: Visual foundation models for sparse self-explainable classifications
por: Turbé, Hugues, et al.
Publicado: (2024)

LL-ViT: Edge Deployable Vision Transformers with Look Up Table Neurons
por: Nag, Shashank, et al.
Publicado: (2025)

Layer by layer, module by module: Choose both for optimal OOD probing of ViT
por: Odonnat, Ambroise, et al.
Publicado: (2026)

Unlocking [CLS] Features for Continual Post-Training
por: Yildirim, Murat Onur, et al.
Publicado: (2025)

Unsqueeze [CLS] Bottleneck to Learn Rich Representations
por: Su, Qing, et al.
Publicado: (2024)

Unsupervised Object Localization in the Era of Self-Supervised ViTs: A Survey
por: Siméoni, Oriane, et al.
Publicado: (2023)

LookupViT: Compressing visual information to a limited number of tokens
por: Koner, Rajat, et al.
Publicado: (2024)

CLAMP-ViT: Contrastive Data-Free Learning for Adaptive Post-Training Quantization of ViTs
por: Ramachandran, Akshat, et al.
Publicado: (2024)

GrowTAS: Progressive Expansion from Small to Large Subnets for Efficient ViT Architecture Search
por: Lee, Hyunju, et al.
Publicado: (2025)

ChAda-ViT : Channel Adaptive Attention for Joint Representation Learning of Heterogeneous Microscopy Images
por: Bourriez, Nicolas, et al.
Publicado: (2023)

ViT-MUL: A Baseline Study on Recent Machine Unlearning Methods Applied to Vision Transformers
por: Cho, Ikhyun, et al.
Publicado: (2024)