CLIPure: Purification in Latent Space via CLIP for Adversarially Robust Zero-Shot Classification
cs.CV, cs.AI
Submitted: 2025-02-25
Updated: 2026-08-29
Code: https://github.com/TMLResearchGroup-CAS/CLIPure
Project page: https://stanislavfort.github.io/2021/01/12/OpenAI_CLIP_adversarial_examples.html
Terminology
Sources
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- MeanSparse: Post-Training Robustness Enhancement Through Mean-Centered Feature Sparsification
- MixedNUTS: Training-Free Accuracy-Robustness Balance via Nonlinearly Mixed Classifiers
- Class-Conditioned Transformation for Enhanced Robust Image Classification
- Robust Classification via a Single Diffusion Model
- Stochastic Security: Adversarial Defense Using Long-Run Dynamics of Energy-Based Models
- Fine-Grained Visual Classification of Aircraft
- Understanding Zero-Shot Adversarial Robustness for Large-Scale Models
- Diffusion Models for Adversarial Purification
- Hierarchical Text-Conditional Image Generation with CLIP Latents
- Fixing Data Augmentation to Improve Adversarial Robustness
- Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models
- Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models
- PixelDefend: Leveraging Generative Models to Understand and Defend against Adversarial Examples
- Score-Based Generative Modeling through Stochastic Differential Equations
- EVA-CLIP: Improved Training Techniques for CLIP at Scale
- Simplified Diffusion Schr\"odinger Bridge
- Guided Diffusion Model for Adversarial Purification
- CoCa: Contrastive Captioners are Image-Text Foundation Models
- Wide Residual Networks
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models