The workshop on Robust Artificial Intelligence System Assurance (RAISA) will focus on research, development and application of robust artificial intelligence (AI) and machine learning (ML) systems. Rather than studying robustness with respect to particular ML algorithms, our approach will be to explore robustness assurance at the system architecture level, during both development and deployment, and within the human-machine teaming context. While the research community is converging on robust solutions for individual AI models in specific scenarios, the problem of evaluating and assuring the robustness of an AI system across its entire life cycle is much more complex. Moreover, the operational context in which AI systems are deployed necessitates consideration of robustness and its relation to principles of fairness, privacy, and explainability.

RAISA’s systems-level perspective will be emphasized via three main thrusts:

• AI System Robustness:  participants will consider techniques for detecting and mitigating vulnerabilities at each of the processing stages of an AI system, including: the input stage of sensing and measurement, the data conditioning stage, during training and application of machine learning algorithms, the human-machine teaming stage, and during operational use.

• The robust development and assured deployment of AI systems: Participants will discuss how to leverage and update common software development paradigms, e.g., DevSecOps, to incorporate relevant aspects of system-level AI assurance.

• The impact of robustness assurance on other AI ethics principles: RAISA will also explore aspects related to ethical AI that overlap and interact with robustness concerns, including security, fairness, privacy, and explainability.

Submission Guidelines

PDF suitable for ArXiv repository (4 to 8 pages). Previously published work (or under-review) is acceptable.

• Submission due date: Friday, November 12, 2021

• Submissions are not anonymized

List of Topics

• AI threat modeling
• AI system robustness
• Explainable AI
• System lifecycle attacks
• System verification and validation
• Robustness benchmarks and standards
• Robustness to black-box and white-box adversarial attacks
• Defenses against training, operational and inversion attacks
• AI system confidentiality, integrity, and availability
• AI system fairness and bias

Committees

Organizing committee

• William Streilein (Chair), MIT Lincoln Laboratory
• Olivia Brown, MIT Lincoln Laboratory
• Rajmonda Caceres, MIT Lincoln Laboratory
• Tina Eliassi-Rad, Northeastern University
• David Martinez, MIT Lincoln Laboratory
• Sanjeev Mohindra, MIT Lincoln Laboratory
• Elham Tabassi, National Institute of Standards and Technology

Venue

The RAISA Workshop is held in conjunction with AAAI-22

February 28 or March 1, 2022

Vancouver, BC, Canada

Contact

All questions about submissions should be emailed to: RAISA@ll.mit.edu