Skip to content

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

Oct 12 · Monday

All papers

Found 10 papers
Filter4

Attack type

Extraction and theft10

Defense type

Model hardening4

Impact

Data and privacy leaks10

System type

Models and APIs9

Attack surface / entry point

User input45Retrieval corpus3Images, audio and video4Training data6Weights and gradients10Inference services and hardware8

Layer

Training and data layer7

System component

Multimodal4
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.
Attack type：Extraction and theftImpact：Data and privacy leaksAttack surface / entry point：Images, audio and videoAttack surface / entry point：Training dataClear all

Another 233papers do not have category tags yet and are not included in filtered results.

  1. DefensesarXiv 2610.06636

    差分隐私混合公共数据集提升隐私学习效果

    提出 DP-MixMin，私有学习公开数据混合并做差分隐私微调

    arXiv
    2610.06636
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    ResNet-18, GPT-2 Medium
    AbstractDP-MixMin 私下学公开混合再 DP 微调，胸片 macro AUC 与邮件困惑度优于均匀混合。

    DP-MixMin 是一条端到端的差分隐私训练流程：先私下决定多个公开数据集的预训练混合，再预训练并在敏感数据上 DP 微调。作者称这是首个这样的流程。

    Paper analysisRead analysis
  2. DefensesarXiv 2610.04426

    UnAct：无需梯度的定向激活干预实现类别遗忘

    提出 UnAct，仅以前向激活干预从分类器中遗忘指定类别

    arXiv
    2610.04426
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    ResNet-18, ViT-B/16
    AbstractUnAct 用前向激活做类别遗忘。

    UnAct 是一种事后类别遗忘：删除请求可以只有少量该类图像，且没有标签和保留数据。它不用梯度，只对遗忘图像做前向。编辑对象是分类器所读的后期单元。

    Paper analysisRead analysis
  4. AttacksarXiv 2610.01365

    ReGap：无需真实隐私数据即可通过微调重新提取模型隐私关联

    提出 ReGap，用自生成伪监督微调恢复模型中的隐私关联

    arXiv
    2610.01365
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Attacker
    White-box, Gray-box
    Evaluated models
    GPT-2 Small, GPT-2 Medium, GPT-2 Large and others6

    Abstract提出无真实私有监督的恢复攻击 ReGap，实验显示微调可通过自生成数据唤醒模型潜在隐私关联，突显后适应隐私风险。

    Paper analysisRead analysis
  5. DefensesarXiv 2609.37858

    存储不等于策略：面向 LLM 遗忘的状态条件支持控制

    提出 Intervention Score 与 DIR-R，按干预效果选择并修订遗忘参数支持

    arXiv
    2609.37858
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Llama 3 8B, Gemma 2 2B, Qwen2.5-7B and others4
    Abstract作者把局部化遗忘拆成初始干预选择和检查点修订，并在三个设置里分别给证据和边界。

    Storage Is Not Strategy 研究局部化 LLM 遗忘里两个常被混在一起的决定：一开始该更新哪些参数，以及优化改变模型之后这个选择要不要改。

    Paper analysisRead analysis
  7. AttacksarXiv 2609.33985

    研究：众包微调数据投毒可放大专有指令抽取

    提出主题锚点投毒，放大众包微调后的专有指令提取

    arXiv
    2609.33985
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Attacker
    Black-box
    Evaluated models
    Qwen2.5-7B-Instruct, Qwen2.5-14B-Instruct, Llama-3.1-8B-Instruct and others4

    Abstract论文提出基于主题锚点的数据中毒攻击，证明低比例的中毒样本可在黑盒条件下隐蔽放大微调指令泄漏。

    Paper analysisRead analysis
  8. Analysis and theoryarXiv 2609.33898

    研究揭示训练效率与模型脆弱性的权衡：高效训练更易受对抗与隐私攻击

    比较高效与高成本训练的对抗和隐私脆弱性并分析损失几何

    arXiv
    2609.33898
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    ResNet-18, Swin Transformer, Qwen2.5-Math-7B-Instruct and others7
    Abstract效用匹配下，三种高效训练都伴随更高的隐私与对抗脆弱性及更尖的损失几何。

    作者在视觉分类和数学推理语言模型上，检查用更少数据或更简对齐换取训练效率时，安全性质是否一起变化。比较在任务效用接近的前提下进行。视觉侧用 KNN-Shapley 的高重要性小子集对照低重要性大子集，并用预训练微调对照从头训练。

    Paper analysisRead analysis
  9. Evaluation and benchmarksarXiv 2609.33481

    IDUnlearn-Bench：面向视觉语言模型的个体级遗忘基准

    提出 IDUnlearn-Bench，评测 VLM 个体级遗忘后的身份残留

    arXiv
    2609.33481
    Published
    Layer
    Training and data layer
    Setting
    Models and APIs
    Evaluated models
    Qwen3-VL-2B, Qwen3-VL-4B, Qwen3-VL-8B and others6
    AbstractIDUnlearn-Bench 用四族探针显示：压低直接属性访问后，身份仍常能被反向查出、绑定或重构。

    IDUnlearn-Bench 评测 VLM 是否忘掉一个可跨模态识别的人，而不只是答不出某条监督事实。

    Paper analysisRead analysis
  10. AttacksarXiv 2508.06837

    Prometheus：针对文生图扩散模型的免训练提示词窃取攻击

    提出 Prometheus，从展示图反推文生图模型的在售提示词

    arXiv
    2508.06837
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    No-box (offline transfer)
    Evaluated models
    Midjourney, Leonardo.ai, DALL·E and others4
    Abstract训练无关的 proxy 搜索攻击。

    Prometheus 是针对文生图市场 showcase 的训练无关提示词窃取：攻击者看不到在售文本，只用本地开源 proxy 恢复 subject 和 modifier，并希望换 subject 后风格仍在。

    Paper analysisRead analysis
You have reached the end