机器遗忘在 ASR 中是否有效：arXiv 论文评估现有算法与评测工具
- arXiv
- 2609.34092
- Published
- Setting
- Models and APIs
- Evaluated models
- E-Branchformer
- DefensesModel hardening
- AttacksExtraction and theft
- ComponentAudio
Abstract梯度上升类方法更接近简单评测下的重训参照，复杂方法会被知情攻击认出，多说话人更差。
All papers
Filter5
System type
Attack surface / entry point
User input51Web pages and interfaces15Documents and messages25Tool responses30Tool and skill descriptions12Code repositories and dependencies4Retrieval corpus11Memory writes40Inter-agent messages6Images, audio and video23Training data3Weights and gradients3Distribution channels9Inference services and hardware3
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.
Another 215papers do not have category tags yet and are not included in filtered results.
- AttacksarXiv 2609.04260
GhostWord：针对自动语音识别的细粒度后门攻击
提出 GhostWord 词级触发后门，对自动语音识别做数据投毒
- arXiv
- 2609.04260
- Published
- Setting
- Models and APIs
- Evaluated models
- Whisper-Small, Whisper-Medium, MMS and others4
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentAudio
- AttacksarXiv 2605.30040
研究揭示 LLM 按 token 计费可被服务商系统性虚报
构造可通过计费审计的 token 虚报，使服务商系统性超收
- arXiv
- 2605.30040
- Published
- Layer
- Infrastructure layer
- Setting
- Models and APIs
- Evaluated models
- PALACE domain-adapted auditor
- AttacksDetection evasion
- TechniqueDefense-targeted
- ComponentMonitors
Abstract不诚实提供方可在三种 token 审计上虚报并通过检查，作者认为需改用提供方不控制的证据。
不透明 LLM 的按 token 计费里，现有审计核验的是提供方可控的证据，不诚实提供方可以虚报并通过检查。。
You have reached the end