seabet app downloadporter:Dr. Yimin Chen
Reporseabet app downloadng locaseabet app downloadon: School Management Building304Room
seabet app download time:2023Yeseabet app download1monseabet app download5Sunday afseabet app downloadrnoon14:00-16:00
seabet app download title: Defending AI Models from Adversarial Attacks through Dseabet app downloadection in Data Space
Personal introducseabet app downloadon:Dr. ChenNowseabet app download of Massachusetts LowellAssistant Professor, School of Computer and Informaseabet app downloadon Science.2018Yeseabet app download Awseabet app downloaddArizona State seabet app downloadPh.D.。The current main research direction is machine learning security and privacy，seabet app download types includeadversarial example seabet app download，backdoor seabet app download，poisoning seabet app download，membership inference seabet app downloadetc.，The application scenario involves federated learning、Comparative learning、Large language model and autonomous driving model。Focus on providing effective and reliable defense systems for machine learning models in practical applications。Many of his reseabet app downloadrch results were published in the Security Conference (IEEE seabet app downloadcurity and Privacy/ACM CCS/ NDSS)He Computer Nseabet app downloadwork Summit (IEEE MobiCOM/INFOCOM）。
Abstract: With the wide application of artificial intelligence models in different fields，The seabet app downloadcurity and robustness of models are becoming increasingly important。Current attacks against artificial intelligence models include the model training phaseabet app downloadpoisoning seabet app download、backdoor seabet app downloadAnd the model deployment phaseabet app downloadadversarial example seabet app download、membership inference seabet app downloadWait.LLMseabet app downloaddAIGCIs greatly expanding the application scenarios of artificial intelligence，The rising model complexity also makes the model have more potential loopholes。In this context，A defense system with controllable costs and a certain degree of versatility is more important。In this seabet app download，We will introduce the team’s two works on the general artificial intelligence model defense mechanism，ieMseabet app downloadDAseabet app downloaddFLAseabet app download。We will focus on how to detect potential seabet app download samples by comparing the different behaviors of samples in data space and representation space，Thus making the corresponding defense system controllable cost，Achieve high detection accuracy for seabet app download methods designed based on different mechanisms。
