TEA-Prompt anchors attention for rehearsal-free continual learning
The method uses 0.38M parameters and stores Gaussian query statistics instead of old images.
Stories per day · 30 days
3 RSS
Relatedattention1autoply-he1cifar-1001continual-learning1dino1dinov31discrete-codes1efficiency1elastic-inference1encrypted-inference1homomorphic-encryption1image-generation1imagenet-r1image-segmentation1multimodal1polynomial-activation1privacy-preserving-ml1prompt-tuning1
18 stories, newest first · one card per story
Stories tagged #vision-transformer
dNaViT encodes images as discrete general-purpose visual tokens
Authors say it matches or beats task-specific baselines on 12 understanding, 4 generation, and one editing benchmark.
Paper trains Vision Transformers to specialize spatial attention heads
Two annotation-free focusing losses steer multi-head attention toward distinct image regions for denser predictions.
AutoPoly-HE: Autonomous Design of Polynomial Activations for Privacy-Preserving Encrypted Inference
AutoPoly-HE automatically designs polynomial activations for encrypted inference across ResNet and Vision Transformer models.
Your Transformer Is Secretly Elastic
A retraining-free method makes pretrained transformers elastic, pruning smoothly to 60% sparsity with small accuracy loss.
One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts
A recurrent vision Transformer matches full-depth accuracy with one block and about 70% fewer parameters.
One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts
A recurrent vision Transformer matches full-depth encoder accuracy with about 70% fewer stored parameters.
TwinViT-DeepJSCC: Adversarially Robust Semantic Image Communication
TwinViT-DeepJSCC defends semantic image links, gaining up to 10.8 dB PSNR under adversarial channel attacks.
TranScope: What the Software Hides About LLM Training Data, the Hardware Reveals at Scale, and Accelerators Magnify
TranScope detects LLM training membership from TLB and accelerator timing without a surrogate model.
Image Classifiers are Efficient Self-Supervised Video Representation Learners
VideoMSN turns pretrained image ViTs into efficient video learners, cutting pretraining by up to 160x.
Copy the Same, Distill the Difference: Initializing Linear Vision Transformers
Copying MLP weights and distilling attention lets linear Vision Transformers match or beat pretrained Softmax ViTs.
ProtoSeam: Lifting Classifier Training with Latent Gaussian Mixture Models
ProtoSeam lifts classifier training with class prototypes, gaining up to five points without changing inference.
VisionHOPE: Visual Backbones as Self-Modifying Learning Systems
VisionHOPE introduces a self-modifying visual backbone where memory content and learning rules co-evolve per image, with competitive ImageNet, COCO, and ADE20K results.
Revisiting Certified Defense with Differential Privacy on Vision Transformers
Laplace PixelDP noise after Vision Transformer patch embeddings collapses accuracy; spectral-norm repair voids the usual certificate.
Reinforcement Learning Inspired Black-box Adversarial Attacks for Computer Vision
A reinforcement-learning black-box attack fools vision models with fewer queries than earlier methods.
Compact Vision Models for Iris Presentation Attack Detection under Presentation Attack Instrument Shift and Environmental Degradation
Compact iris presentation-attack detection models under 0.3M parameters show APCER rising 17-30 percentage points against unseen attack instruments, indicating limited deployment readiness.
Bridging Modalities on the Cortex: Surface-based MRI to PET Translation with a Diffusion Bridge
DB-SUiT is a surface-based diffusion bridge generating synthetic FDG-PET from MRI on the cortical manifold, lifting dementia classification 14.2% over MRI.
Lightweight Vision Transformer Compression for On-Device Plant Disease Detection in Resource-Constrained Agricultural Field Conditions
A unified ViT compression pipeline (H-BAC pruning, quantization, distillation) cuts plant-disease models 54.5x to 6.01 MB while keeping 95.13% accuracy.