Sign Up

Towards Future Deep Learning Models: Representation and Bias

 

Abstract:  In this talk, I will present my ideas on future deep learning systems that ought to be bias-free from human factors. The input to these systems is represented in a format where information is not revealed and should further help achieve fair decision-making.  I will present our experiments on face recognition where some such factors have been examined.  Additionally, I will introduce a new representation scheme for 3D data, reducing storage and improving cybersecurity. I will also present some of our on-going projects that include sea ice motion analysis (https://mosaic-expedition.org/team/partner-institutions/), Augmented Reality for robotic surgical platforms (http://bigdatavision.org/vrsurgery/), and hidden target detection (https://www.army.mil/article/242508).

 

Bio:  Dr. Chandra Kambhamettu is a Full Professor in the Computer Science department at the University of Delaware, where he directs the Video/Image Modeling and Synthesis (VIMS) group. His research interests include computer vision, deep learning, big data visual analytics, biomedical image analysis, bioinformatics, and virtual reality. His Lab focuses on novel schemes for multi-modal image analysis methodologies. Some of his recent research includes deep learning methods for 3D representation, classification, and modeling, augmented reality-based visualization of intra and post-surgical procedures, remote surgery, miniature camera-based 3D reconstruction for medical applications, plant science image analysis, vehicle/ship-mounted cameras (visible and IR) for environment sensing and understanding. Several of Dr. Kambhamettu’s works focused on problems that highly impact earth life, such as arctic sea ice observations with application towards mammal habitat quantification and climate change, hurricane image studies, among several others. His specific interest includes translational research using AI-based image analysis and Virtual Reality (VR) tools. Dr. Kambhamettu has supervised eighteen Ph.D. students, several Master’s and undergraduate theses, and projects in his expertise areas. Before joining UD, he was a research scientist at NASA-Goddard, where he received the “1995 Excellence in Research Award.” He received the NSF CAREER award in 2000 and NASA Group Achievement Award” in 2013 for his work in deploying the Arctic Collaborative Environment. Dr. Kambhamettu was the associate editor for the journals IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Pattern Recognition Letters, Pattern recognition, and special issue guest editor, Image and Vision Computing journal. He organized IEEE workshops on Big Data Computer Vision, Mathematical Methods on Biomedical Image Analysis, Articulated and Nonrigid Motion analysis, and Undergraduate education in computer vision. HeVwas co-editor of a special issue on “Content Analysis for Big Multimedia Data” in Springer’s Multimedia Tools and Applications journal and co-chair of an “International Workshop on Visual Odometry and Computer Vision Applications Based on Location Clues,” held in conjunction with IEEE Computer Vision and Pattern Recognition (CVPR) conference (2017-2021). He is currently involved in building ice motion products as part of the MOSAiC expedition (https://mosaic-expedition.org/).

Event Details

See Who Is Interested

0 people are interested in this event


User Activity

No recent activity