Semantic-aware Adversarial Fine-tuning for CLIP

• URL: http://arxiv.org/abs/2602.12461v1
• Date: Thu, 12 Feb 2026 22:40:42 GMT
• Title: Semantic-aware Adversarial Fine-tuning for CLIP
• Abstract summary: It has been shown that cosine similarity between a single image and a single hand-crafted template is insufficient to measure the similarity for image-text pairs.<n>We propose Semantic-aware Adversarial Fine-Tuning (SAFT), which fine-tunes CLIP's image encoder with semantic-aware AEs.
• Score: 32.894711409751956
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent studies have shown that CLIP model's adversarial robustness in zero-shot classification tasks can be enhanced by adversarially fine-tuning its image encoder with adversarial examples (AEs), which are generated by minimizing the cosine similarity between images and a hand-crafted template (e.g., ''A photo of a {label}''). However, it has been shown that the cosine similarity between a single image and a single hand-crafted template is insufficient to measure the similarity for image-text pairs. Building on this, in this paper, we find that the AEs generated using cosine similarity may fail to fool CLIP when the similarity metric is replaced with semantically enriched alternatives, making the image encoder fine-tuned with these AEs less robust. To overcome this issue, we first propose a semantic-ensemble attack to generate semantic-aware AEs by minimizing the average similarity between the original image and an ensemble of refined textual descriptions. These descriptions are initially generated by a foundation model to capture core semantic features beyond hand-crafted templates and are then refined to reduce hallucinations. To this end, we propose Semantic-aware Adversarial Fine-Tuning (SAFT), which fine-tunes CLIP's image encoder with semantic-aware AEs. Extensive experiments show that SAFT outperforms current methods, achieving substantial improvements in zero-shot adversarial robustness across 16 datasets. Our code is available at: https://github.com/tmlr-group/SAFT.

Related papers

• Measuring Similarity between Artistic and AI Generated Images using Siamese Neural Networks [0.06524460254566902]
  This research quantifies the similarity between original pieces and AI-generated counterparts.<n>We use Siamese Networks with frozen CLIP encoders and cosine similarity optimized through triplet loss.
  arXiv  Detail & Related papers  (2026-08-24T19:42:35Z)
• Unleashing Degradation-Carrying Features in Symmetric U-Net: Simpler and Stronger Baselines for All-in-One Image Restoration [52.82397287366076]
  All-in-one image restoration aims to handle diverse degradations (e.g., noise, blur, adverse weather) within a unified framework.<n>In this work, we reveal a critical insight: well-crafted feature extraction inherently encodes degradation-carrying information.<n>Our symmetric design preserves intrinsic degradation signals robustly, rendering simple additive fusion in skip connections.
  arXiv  Detail & Related papers  (2025-12-11T12:20:31Z)
• When Semantics Regulate: Rethinking Patch Shuffle and Internal Bias for Generated Image Detection with CLIP [13.360123625878733]
  CLIP-based detectors often rely on semantic cues rather than generator artifacts, leading to brittle performance under distribution shifts.<n>We show that Patch Shuffle provides an unusually strong benefit for CLIP, that disrupts global semantic continuity.<n>We propose SemAnti, a semantic-antagonistic fine-tuning paradigm that freezes the semantic subspace and adapts only artifact-sensitive layers.
  arXiv  Detail & Related papers  (2025-11-24T13:54:00Z)
• Beyond Cosine Similarity Magnitude-Aware CLIP for No-Reference Image Quality Assessment [25.104682483704]
  We introduce a novel adaptive fusion framework that complements cosine similarity with a magnitude-aware quality cue.<n>Our method consistently outperforms standard CLIP-based IQA and state-of-the-art baselines, without any task-specific training.
  arXiv  Detail & Related papers  (2025-11-13T04:28:03Z)
• Semantic Visual Anomaly Detection and Reasoning in AI-Generated Images [96.43608872116347]
  AnomReason is a large-scale benchmark with structured annotations as quadruple textbfAnomAgent<n>AnomReason and AnomAgent serve as a foundation for measuring and improving the semantic plausibility of AI-generated images.
  arXiv  Detail & Related papers  (2025-10-11T14:09:24Z)
• Semantic-Aware Reconstruction Error for Detecting AI-Generated Images [22.83053631078616]
  We propose a novel representation, namely Semantic-Aware Reconstruction Error (SARE), that measures the semantic difference between an image and its caption-guided reconstruction.<n>SARE provides a robust and discriminative feature for detecting fake images across diverse generative models.<n>We also introduce a fusion module that integrates SARE into the backbone detector via a cross-attention mechanism.
  arXiv  Detail & Related papers  (2025-08-13T04:37:36Z)
• Unleashing Text-to-Image Diffusion Prior for Zero-Shot Image Captioning [70.98890307376548]
  We propose a novel Patch-wise Cross-modal feature Mix-up (PCM) mechanism to adaptively mitigate the unfaithful contents during training.<n>Our PCM-Net ranks first in both in-domain and cross-domain zero-shot image captioning.
  arXiv  Detail & Related papers  (2024-12-31T13:39:08Z)
• Wavelet-based Unsupervised Label-to-Image Translation [9.339522647331334]
  We propose a new Unsupervised paradigm for SIS (USIS) that makes use of a self-supervised segmentation loss and whole image wavelet based discrimination. We test our methodology on 3 challenging datasets and demonstrate its ability to bridge the performance gap between paired and unpaired models.
  arXiv  Detail & Related papers  (2023-05-16T17:48:44Z)
• Semantic Image Synthesis via Diffusion Models [174.24523061460704]
  Denoising Diffusion Probabilistic Models (DDPMs) have achieved remarkable success in various image generation tasks.<n>Recent work on semantic image synthesis mainly follows the de facto GAN-based approaches.<n>We propose a novel framework based on DDPM for semantic image synthesis.
  arXiv  Detail & Related papers  (2022-06-30T18:31:51Z)
• Introspective Deep Metric Learning for Image Retrieval [80.29866561553483]
  We argue that a good similarity model should consider the semantic discrepancies with caution to better deal with ambiguous images for more robust training. We propose to represent an image using not only a semantic embedding but also an accompanying uncertainty embedding, which describes the semantic characteristics and ambiguity of an image, respectively. The proposed IDML framework improves the performance of deep metric learning through uncertainty modeling and attains state-of-the-art results on the widely used CUB-200-2011, Cars196, and Stanford Online Products datasets.
  arXiv  Detail & Related papers  (2022-05-09T17:51:44Z)
• No Token Left Behind: Explainability-Aided Image Classification and Generation [79.4957965474334]
  We present a novel explainability-based approach, which adds a loss term to ensure that CLIP focuses on all relevant semantic parts of the input. Our method yields an improvement in the recognition rate, without additional training or fine-tuning.
  arXiv  Detail & Related papers  (2022-04-11T07:16:39Z)
• USIS: Unsupervised Semantic Image Synthesis [9.613134538472801]
  We propose a new Unsupervised paradigm for Semantic Image Synthesis (USIS) USIS learns to output images with visually separable semantic classes using a self-supervised segmentation loss. In order to match the color and texture distribution of real images without losing high-frequency information, we propose to use whole image wavelet-based discrimination.
  arXiv  Detail & Related papers  (2021-09-29T20:48:41Z)
• Causal Intervention for Weakly-Supervised Semantic Segmentation [122.1846968696862]
  We aim to generate better pixel-level pseudo-masks by using only image-level labels. We propose a structural causal model to analyze the causalities among images, contexts, and class labels. Based on it, we develop a new method: Context Adjustment (CONTA), to remove the confounding bias in image-level classification.
  arXiv  Detail & Related papers  (2020-09-26T09:26:29Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.