[Eecs_mscs] EECS Research Seminar Talk-Towards Trustworthy AI: Evolving from Secure Systems to Responsible Integration

• Previous message: [Eecs_mscs] EECS Research Seminar Talk-Advancing Trusted and Resilient Computing: ML-Based Cyberattacks Detection, Hardware Vulnerability Assessment, and PUF Authentication for IoT Security
• Next message: [Eecs_mscs] EECS Research Seminar Talk-Run-Time Analysis and Security of Multi-Language Systems
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Title:
Towards Trustworthy AI: Evolving from Secure Systems to Responsible Integration

Abstract:
Trustworthy Artificial Intelligence (AI) has become a significant concern as machine learning (ML) models become integral to safety-critical systems like vehicle autopilots and medical diagnostics. Consequently, the focus extends beyond simple effectiveness and efficiency to also encompass the resilience of these systems against errors and adversarial attacks. Within the spectrum of such attacks, poisoning attacks represent a significant threat, compromising the integrity of models by inserting malicious instances into their training sets.
In this presentation, I will share my research contributions to machine learning security, specifically focusing on poisoning attacks and defenses: (1) improving ML model robustness against untargeted poisoning attacks through an unsupervised anomaly detection method grounded in the Bayesian Information Criterion; and (2) investigating the vulnerabilities of ML models in novel applications, such as video action recognition systems, to carefully designed targeted poisoning attacks.
Furthermore, I will outline the direction of my future research on Trustworthy AI. This includes aligning my current research in ML security with the emerging trend of foundation models and broadening my research to incorporate ethical and privacy considerations within AI. I aim to support technological advancement and address societal challenges through reliable AI.

Biography:
Xi Li is currently a Ph.D. candidate in the Department of Computer Science and Engineering at Penn State University, supervised by Dr. George Kesidis and Dr. David Miller. She received her B.S. degree in Electrical Engineering from Southeast University, Nanjing, China, in 2016, and her M.S. degree in Computer Science from Penn State University in 2018.
Her research interests include trustworthy AI and adversarial machine learning, with her Ph.D. thesis specifically focusing on poisoning attacks and defenses against deep neural networks. Her research vision is centered on developing trustworthy and reliable AI systems, aiming to support the advancement of technology and solve social challenges.
She has published papers at conferences such as ICCV, AAAI, and ICASSP, and in journals like TKDE.
You can find out more about Xi Li at https://lixi1994.github.io/<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flixi1994.github.io%2F&data=05%7C02%7Ceecs_mscs%40listserv.ohio.edu%7C70d40dad9f1743162bd808dc32348eb4%7Cf3308007477c4a70888934611817c55a%7C0%7C0%7C638440448534980154%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=lsNXJt4gjkLOF%2BuWNDuiHIzprWZPVQqlFjHGs1WYnXw%3D&reserved=0 
>

________________________________________________________________________________
Microsoft Teams meeting
Join on your computer, mobile app or room device
Click here to join the meeting< https://teams.microsoft.com/l/meetup-join/19%3ameeting_ODc0Mzk0OWItY2ZkYy00NjRmLWEwMWUtNmRkYmUzNmQ3YjM3%40thread.v2/0?context=%7b%22Tid%22%3a%22f3308007-477c-4a70-8889-34611817c55a%22%2c%22Oid%22%3a%22685c3f4f-29d5-4141-ada5-0fdeab8480e4%22%7d 
>
Meeting ID: 260 784 379 62
Passcode: Rqgg4w
Download Teams< https://www.microsoft.com/en-us/microsoft-teams/download-app 
> | Join on the web< https://www.microsoft.com/microsoft-teams/join-a-meeting 
>
Or call in (audio only)
+1 614-706-6572,,324005943#<tel:+16147066572,,324005943#>   United States, Columbus
Phone Conference ID: 324 005 943#
Find a local number< https://dialin.teams.microsoft.com/8f5f7319-0053-4423-a154-4f8b6e7fb7dd?id=324005943 
> | Reset PIN< https://dialin.teams.microsoft.com/usp/pstnconferencing 
>
[ https://www.ohio.edu/sites/default/files/2018-11/invite_logo_teams.jpg 
]
If you encounter issues with this meeting, please visit the Help link. If you are not able to resolve the problems, please contact the meeting organizer to let them know you are having difficulty.
Learn More< https://aka.ms/JoinTeamsMeeting 
> | Help< https://www.ohio.edu/oit/services/collaboration/teams/help 
> | Meeting options< https://teams.microsoft.com/meetingOptions/?organizerId=685c3f4f-29d5-4141-ada5-0fdeab8480e4&tenantId=f3308007-477c-4a70-8889-34611817c55a&threadId=19_meeting_ODc0Mzk0OWItY2ZkYy00NjRmLWEwMWUtNmRkYmUzNmQ3YjM3@thread.v2&messageId=0&language=en-US 
>
________________________________________________________________________________

-------------- next part --------------
An HTML attachment was scrubbed...
URL: < http://listserv.ohio.edu/pipermail/eecs_mscs/attachments/20240220/a9c2f886/attachment.html 
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/calendar
Size: 11125 bytes
Desc: not available
URL: < http://listserv.ohio.edu/pipermail/eecs_mscs/attachments/20240220/a9c2f886/attachment.ics 
>

• Previous message: [Eecs_mscs] EECS Research Seminar Talk-Advancing Trusted and Resilient Computing: ML-Based Cyberattacks Detection, Hardware Vulnerability Assessment, and PUF Authentication for IoT Security
• Next message: [Eecs_mscs] EECS Research Seminar Talk-Run-Time Analysis and Security of Multi-Language Systems
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]