The AI Assurance Workshop leadership team will meet to share information about their work and to plan a full Workshop on AI Assurance for Spring 2021.
This workshop discussion will focus on ‘How do we test AI/ML performance, including assessment of function, safety, and security?’ Participants will give brief presentations and entertain questions. Finally, as a group, we will draft the agenda for a Spring 2021 workshop.
Contacts: Edward Griffor (edward.griffor@nist.gov) and Marcello Balduccini (mbalducc@sju.edu).
The meeting will be a virtual, using Go2Meeting (see invitation for the link)
There is no cost associated with joining the meeting
Topics will include: (slides)
Measuring AI performance
AI Ethics
AI-enabled Cyber-Physical Systems/IoT
Adversarial AI
UQ/Trust for AI
Trusted Hardware/CPS
Explainable AI for CPS
Physics Informed AI
Statistical Learning/UQ for AI
Chris Greer (NIST) Welcome (5 min)
Ed Griffor (NIST) Purpose and Scope (10 min) (slides)
Marcello Balduccini (SJU) Topics and CPS/IoT (10 min) (slides)
John Feddema (Sandia) Sandia’s Trusted AI initiative 10 min (slides)
Jeremy Wendt (Sandia) ML security concerns in the counter-adversarial space (10min + 5min Q&A) (slides)
Linyu Lin, Nam Dinh (NC State) Nearly autonomous management and control of nuclear reactors (15min + 5min Q&A) (slides)
Nicholas Mattei (Tulane) AI Ethics Overview (10min + 5min Q&A) (slides)
V.S. Subrahmanian (Dartmouth) AI-Enabled Prediction of Cyber-Attacks, and AI-Enabled Cyber-Deception (10min + 5min Q&A) (slides)
Elham Tabassi (NIST) NIST Trustworthy AI program (10min + 5min Q&A) (slides)
Michael Reed Smith (Sandia) Model credibility, mathematically rigorous trust models and explanations (10min + 5min Q&A) (slides)
Chasity Nadeau (SJU) XAI and CPS: possibilities and challenges (10min + 5min Q&A) (slides)
DRAFT Statement of Purpose: Assessing the need for a consensus AI Assurance measurement/testing methodology
Workshop Topics
Speaker List (Suggestions/Recommendations)