Skip to content

Translation in progress Some entries still include text in the original language.

Oct 13 · Tuesday

All papers

Found 3 papers
Filter5

System type

Models and APIs3

Attacker access

Black-box5White-box3

Attack surface / entry point

Images, audio and video3

Layer

Model layer3Application layer0

System component

Fine-tuning and open weights3Vision3Audio2Embeddings1Multimodal7
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 215papers have no tags yet and are not included in filtered results.
Attacker access：White-boxLayer：Model layerLayer：Application layerSystem component：AudioSystem component：EmbeddingsClear all

Another 215papers do not have category tags yet and are not included in filtered results.

  3. DefensesarXiv 2607.14737

    GeoDetect：面向视觉语言预训练模型的几何对抗检测方法

    提出 GeoDetect 以几何分数检测视觉语言预训练模型的对抗样本

    arXiv
    2607.14737
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box
    Evaluated models
    CLIPViT (ViT-B/16), CLIPCNN (ResNet-50), ALBEF and others4
    AbstractGeoDetect 用几何分数检测 VLP 对抗样本。

    GeoDetect 是面向视觉-语言预训练模型的对抗样本检测方法，用嵌入几何而不是任务 logits 区分干净样本与对抗样本。

    Paper analysisRead analysis
You have reached the end