Educational Games and Debuggers for Machine Learning
Abstract
The PIs, Dr. Randall Davis and Dr. Henry Lieberman, have a novel set of research techniques to play with predefined machine learning models to provide learning-training and testing to a system. The PIs plan to focus on machine learning performed by non-expert persons rather than scientists-engineers, real end-users. The object is to teach a simulated autonomous car to drive itself. They will study the design of a system called Learner s Permit that uses the easily comprehensible interface metaphors of visual rules and Programming by Demonstration techniques to provide a concrete visual representation that users can manipulate to understand and control machine learning programs. They aim to support the entire ML workflow, including data collection, training, testing, model criticism and debugging. Though they target beginning users to promote education; to keep themselves honest about comprehensibility, their principles will also enable more productive programming environments for expert professional AI developers.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 14, 2024
- Source ID
- FA95502310744
Entities
People
- Randall Davis
Organizations
- Air Force Office of Scientific Research
- Massachusetts Institute of Technology
- United States Air Force