Is It a Wolf or a Husky? Can You Trust Your AI Vision Model?

Artificial intelligence is ubiquitous in computer vision applications - from self-driving cars to retail, healthcare, manufacturing, and beyond. How do you know your computer vision AI model is working as advertised? Way covers various ways that AI models can be fooled, such as by adding stickers to a stop sign. He focuses on using techniques to probe how a vision AI model makes a decision. Finally, Way shows how the LIME (local interpretable model-agnostic explanations) technique explains what an AI model is doing when it's looking at pictures of wolves and huskies, for example.

About the presenter
Ted Way, Ph.D., is a program manager lead for the Microsoft Insights Apps AI team. He is passionate about telling the story of how AI will empower people and organizations to achieve more. He has bachelor's degrees in electrical engineering and computer engineering, master's degrees in electrical engineering and biomedical engineering, and a doctorate in biomedical engineering from the University of Michigan in Ann Arbor. He wrote his doctoral dissertation on "spell check for radiologists," a computer-aided diagnosis (CAD) system that uses image processing and machine learning to predict lung cancer malignancy on chest CT scans.