Matthew Brown, PhD
Professor and Director, UCLA Center for Computer Vision and Imaging Biomarkers (CVIB)
Embedding deep neural networks within Cognitive AI for machine reasoning and automatic parameter tuning
This half-day MICCAI 2022 (September 22, 2022, 8:00am GMT+8, Singapore) tutorial will introduce the concept of Cognitive AI and how it can be used to address the limitations of current Narrow AI, especially in translation of medical imaging AI into clinical practice. We will introduce practical techniques to support deep neural networks with machine reasoning and an open-source software framework for implementation of Cognitive AI.
Attendees will learn how to use Cognitive AI techniques to make deep neural networks more reliable, mitigate failures, and deliver trustworthy, explainable AI in clinical practice. A live, hands-on demonstration will give attendees a quick start in using the SimpleMind framework to complement their current developments in AI. The framework enhances deep neural networks with knowledge-based machine reasoning, general-purpose pre/post processing agents, and automatic end-to-end parameter optimization.
We encourage attendees to review a more detailed description of the tutorial, as well as details of each talk and hands-on component. Attendees are also encouraged to review the Git repository before attending the tutorial.
Learning Objectives
- Understand the limitations of Narrow AI in medical imaging and the potential of Cognitive AI to address them.
- Learn how human level knowledge of image content and agent-based processing can be represented within a model and applied during image segmentation.
- Understand approaches to automatic, end-to-end optimization of all system parameters and hyper parameters, including grid search and genetic algorithms.
- Gain insights from case studies on how a Cognitive AI approach can be applied in medical image segmentation to improve the accuracy and reliability of DNNs.
- Receive hands-on training using Colab notebooks to easily download, configure, and use an open-source framework that embeds DNNs within Cognitive AI.
Agenda
| Didactic Session | ||
|---|---|---|
| 08:00-08:10 | Welcome & Introduction | Matthew Brown, PhD |
| 08:10-08:30 | Brittleness of Neural Networks & Explainable AI | William Hsu, PhD |
| 08:30-08:50 | Medical AI Meets Reality | Jonathan Goldin, MD PhD |
| 08:50-09:20 | Keynote: A New Cognitive AI Paradigm - From Neural Networks to SimpleMind | Matthew Brown, PhD |
| 09:20-09:40 | Machine Reasoning: A Practical Implementation | Youngwon Choi, PhD |
| 09:40-09:50 | Coffee Break | |
| 09:50-10:10 | A Virtual Data Scientist: Automatic Parameter Tuning | M. Wasil Wahi-Anwar, MS |
| SimpleMind Framework: Hands-On Workshop | Youngwon Choi, M. Wasil Wahi-Anwar, John Hoffman | |
|---|---|---|
| 10:10-10:25 | Code Pull & Install | |
| 10:25-10:45 | Creating a SM Model and Machine Reasoning | |
| 10:45-11:05 | APT Monitoring and SM Result Analysis | |
| 11:05-11:30 | Discussion / Q&A | |
Organizing Team
Youngwon Choi, PhD
Postdoctoral Fellow, UCLA Computer Vision and Imaging Biomarkers (CVIB)
Jonathan Goldin, MD PhD
Professor and Executive Vice Chair, UCLA Radiology
William Hsu, PhD
Associate Professor of Radiology and Bioengineering, UCLA Medical & Imaging Informatics
John Hoffman, PhD
Adjunct Faculty, UCLA Computer Vision and Imaging Biomarkers (CVIB)
M. Wasil Wahi-Anwar, MS
PhD Candidate, UCLA Computer Vision and Imaging Biomarkers (CVIB)
Contact Us
Curious to learn more before the tutorial? Watch our informational YouTube video on SimpleMind AI and its capabilities, or contact us. We also welcome discussion about any and all collaborations!
