June 5, 10:15am
Location: Santa Clara I

MLOps Automation for the Capture and Analysis of Responsible AI Metrics

co-presented with Ramesh Menon

The United States Defense Intelligence Agency Chief Technology Office (DIA/CTO) is piloting Responsible Artificial Intelligence in Machine Learning (AI/ML) by exercising Department of Defense (DoD) and United States policy upon a reference AI/ML architecture.  This architecture – consistent with U.S. values – will be used to facilitate the creation of processes, tools, and training to ensure ethical, reduced bias, and data secure AI/ML that are essential for fostering a positive effect on society. Our team will introduce this architecture to perform natural language classification with the goal of identifying areas to increase trust and acceptance of AI/ML generated outputs in the context of Human-AI teaming. A key element will be to promote explainability of the inferencing of the AI/ML allowing introspection into metrics from data bias analysis, model analysis, and performance metrics as these relate to AI/ML Strategy, Policies, and constraints for the Responsible use of AI/ML. 

Bruce Goldfeder, M.A., is the Senior Engineering and Technical Advisor to the Defense Intelligence Agency (DIA) and Chief Technology Officer (CTO). As the Senior Engineering and Technical Advisor, he provides subject matter expertise in the field of artificial intelligence and machine learning (AI/ML).  Mr. Goldfeder serves on the DIA CTO Emerging Technology team providing AI/ML prototyping, technology augmentation for ongoing AI/ML programs, and future technology solutions advancing DIA mission needs. Mr. Goldfeder received his bachelor’s degree in Electrical Engineering from Lehigh University, Master’s degree in Computer Science from The George Washington University, and is currently a PhD Candidate in the George Mason University Computations Sciences and Informatics program.  Mr. Goldfeder, a former Air Force Officer, has worked for over 25 years as a contractor supporting multiple Department of Defense and Intelligence Community agencies, and the Defense Advanced Research Projects Agency.