HBU2022 @ ICPR



IAPR

Microsoft

ChaLearn

Keynotes


Bridging Fingers - Gestures for Knowledge Gain

Prof. Juan Wachs (Purdue University)

Abstract: In this talk, I will discuss the importance of natural physical expression as means for interacting with devices, machines and robots in the healthcare domain. While the main discussion focus is on using gesture interaction, the area of embodiment will be explored as well. With the recent advent of commodity sensors, like the Kinect, Wii, Leap Motion and MYO arm bands, exciting and provocative directions of research and applications are gaining traction among the industry and academic communities. In this talk, I will discuss insights and findings about the use of gestures to control robots for collaboration with surgeons, for surgical training and for rehabilitation.

Bio: Juan P. Wachs received the B.Ed.Tech. degree in electrical education with ORT Academic College, Hebrew University of Jerusalem, Jerusalem, Israel, and the M.Sc. and Ph.D. degrees in industrial engineering and management from the Ben-Gurion University of the Negev, Beersheba, Israel. He is an Associate Professor with the Industrial Engineering School, Purdue University, West Lafayette, IN, USA. He is the Director of the Intelligent Systems and Assistive Technologies Laboratory, Purdue University, and he is affiliated with the Regenstrief Center for Healthcare Engineering. He completed Post-Doctoral training with the Naval Postgraduate School's MOVES Institute under a National Research Council Fellowship from the National Academies of Sciences. Dr. Wachs was a recipient of the 2013 Air Force Young Investigator Award, the 2015 Helmsley Senior Scientist Fellow, and he is a 2016 Fulbright Scholar. He is also the Associate Editor of the IEEE Transactions in Human-Machine Systems and Frontiers in Robotics and AI.

Should we deploy our research?

Prof. Ehsan Hoque (University of Rochester)

Abstract: Deploying our research in the real world to collect and validate behavioral data from humans is labor intensive. The cycle requires designing a protocol, getting approval from IRB, building the data collection infrastructure, workforce to support the continuous data collection process, and ensuring the diversity and integrity of data. After all this incredible (and not publishable) amount of effort, the data could be incomplete, noisy, and mostly unusable. Should academics even worry about deploying their work or continue to push the algorithmic boundaries using available data? In this talk, I will share our years of experience deploying work in the real-world setting, from allowing people to practice public speaking to individuals with movement disorders performing neurological tests - using a computer browser and a webcam. I will highlight some of the 'accidental findings' through deployment and how they led to new scientific discoveries and research opportunities. The talk will provide guidelines on making the call on deployment in academic research and translating challenges into new opportunities.

Bio: Ehsan Hoque is an associate professor of computer science at the University of Rochester, where he co-directs the Rochester Human-Computer Interaction (ROC HCI) Lab. From January 2018 to June 2019, he was the interim director for the Goergen Institute for Data Science. He co-leads the Rochester Human-Computer Interaction (ROC HCI) Group. He received his PhD from Massachusetts Institute of Technology in 2013. His research program aims to use techniques from artificial intelligence to amplify human ability. He models and captures the dynamics of human behavior and their relationships using machine learning, computer vision, and network sciences, and design interactive systems to promote equity and access in health care and education. He is the recipient of a NSF Career Award (2018), and was an MIT Top 35 Innovator under 35 (2016). He served as an Associate Editor of IEEE Transactions on Affective Computing (2015-2019), PACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT) (2017-2020), and Digital Biomarkers. He was a program chair of ISBA 2017.

 

Using Interpretable Machine Learning to Understand Clinical Behavior and Optimize Healthcare

Dr. Rich Caruana (Microsoft Research)

Abstract: Clinicians are human just like everyone else. Because of this, they sometimes make suboptimal decisions or exhibit bias. We have developed a glass-box machine learning method that is as accurate as black-box methods such as deep neural nets, boosted trees and random forests on tabular clinical data, and yet which is fully interpretable. Using this model on clinical datasets has revealed a wealth of information about how clinicians make decisions. It has also suggested ways in which clinical decision making might be improved. Glass-box models trained on clinical data also demonstrate the risk of relying on black-box models in healthcare --- all clinical data has surprising flaws that cause models trained on them to be potentially risky. In the presentation we’ll talk about human clinical decision making, surprises that are lurking in clinical data, using glass-box machine learning to optimize healthcare delivery, and methods for protecting privacy and detecting bias.

Bio: Rich Caruana is a senior principal researcher at Microsoft Research. Before joining Microsoft, Rich was on the faculty in the Computer Science Department at Cornell University, at UCLA's Medical School, and at CMU's Center for Learning and Discovery. Rich's Ph.D. is from Carnegie Mellon University, where he worked with Tom Mitchell and Herb Simon. His thesis on Multi-Task Learning helped create interest in a new subfield of machine learning called Transfer Learning. Rich received an NSF CAREER Award in 2004, best paper awards in 2005 (with Alex Niculescu-Mizi), 2007 (with Daria Sorokina), and 2014 (with Todd Kulesza, Saleema Amershi, Danyel Fisher, and Denis Charles), and co-chaired KDD in 2007. His current research focus is on learning for medical decision making, transparent modeling, and deep learning for weather forecasting.