CMU SEI Research Review 2021 Day 3 Artifacts
• Collection
Publisher
Software Engineering Institute
Subjects
Abstract
This collection includes slides and videos from Day 3 of the 2021 SEI Research Review event. These sessions include presentations, lightning talks, and collaboration conversations. Day 3 sessions include Collaboration Conversation on Human-Centered AI, Train but Verify: Towards Practical AI Robustness, Knowing When You Don't Know: AI Engineering in an Uncertain World, Predicting Inference Degradation in Production ML Systems, Collaboration Conversation on AI Fusion and AI Engineering: Enabling AI in Complex Environments, and Collaboration Conversation on the Implementation of the National Agenda for Software.
Collection Items
Train but Verify: Towards Practical AI Robustness
• Presentation
By Nathan M. VanHoudnos
Train, but Verify is an effort to develop an AI Engineering process to train AI systems to have specific robustness properties.
Learn More
Knowing When You Don’t Know: AI Engineering in an Uncertain World
• Presentation
By Eric Heim
Presents a method to improve AI system robustness that evaluates machine-learned classifiers and derives metrics that directly measure calibration performance.
Learn More
Predicting Inference Degradation in Production ML Systems
• Presentation
By Grace Lewis
Proposes developing empirically validated metrics and a test harness to predict a model's inference quality degradation due to different types of data drift.
Learn MorePart of a Collection
CMU SEI Research Review 2021