IEEE SMC Victorian Chapter Lecture on Resilient and Safe AI

SafeFM: Advancing Safe Foundation Models via Cross-Modal Attacks and Visual-Resampling-inspired Guards

Speaker: 

Dr. Qing Guo, an assistant professor at the National University of Singapore (NUS) and a Senior Research Scientist, A*STAR in Singapore.

Teams meeting:

On-line meeting link

Friday, 18 Oct 2024 at 12.00am

IEEE Victorian Section SMC Chapter. Organized by Prof. Saeid Nahavandi and A/Prof. Hailing Zhou.

Abstract:
Over the past two years, we have witnessed the transformative potential of implementing state-of-the-art artificial intelligence (AI) technologies in real-world applications, driven by the remarkable outcomes of foundation models (FMs) combined with innovative business strategies. Generally, an FM refers to any model trained on extensive data that can be fine-tuned for various downstream tasks, including CLIP, SAM, ChatGPT, BLIP, and stable diffusion models. With the novel business strategy, ordinary individuals can leverage these tools to generate personalized content with tailored prompts, which inevitably raises significant security concerns. To ensure the safety of foundation models, we focus on two key directions: exploring adversarial attacks against FMs to uncover vulnerabilities in state-of-the-art models and developing effective defense mechanisms for enhancement. In this presentation, we will introduce our recent research efforts in attacking FMs, including visual language models and diffusion models, from various perspectives such as transferability and efficiency. Additionally, we will delve into our recent work on leveraging visual resampling principles to defend against potential attacks, thereby enhancing the reliability of FMs without altering their model weights.

Biography: 
Dr. Qing Guo is currently an adjunct assistant professor at the National University of Singapore (NUS). He is also a Senior Research Scientist and principal investigator (PI) at the Centre for Frontier AI Research (CFAR), Agency for Science, Technology, and Research (A*STAR) in Singapore. In 2019, he joined Nanyang Technological University (NTU) as a Research Fellow and was appointed as a Wallenberg-NTU Presidential Postdoctoral Fellow in 2020. He received the Best Platinum Paper Award at ICME in 2018, the ACM Tianjin Outstanding Doctoral Dissertation Award in 2020, the third place in the AISG Trusted Media Challenge 2022, won the Best Paper Award at the ECCV 2022 AROW workshop, awarded AISG Robust AI Grant Challenge in 2023 and the Digital Trust Centre Research Grant in 2024. His research mainly focuses on AI safety and computer vision. He has published over 50 papers in top-tier conferences and journals. He serves as a Senior PC for AAAI 2023/2024 and vertical chair for Resilient and Safe AI at the IEEE Conference on Artificial Intelligence (CAI) in 2024.