DisCoMath Seminar: Relaxing Balanced Graph Cuts, Part II
Relaxing Balanced Graph Cuts, Part II
Dr. Nathan Cahill
Associate Professor
Director, Mathematical Modeling Ph.D. Program
School of Mathematical Sciences, RIT
Abstract:
In April, I showed how image segmentation can be thought of as a process of partitioning an image into a small number of subsets, and how image segmentation algorithms can often be modeled as graph partitioning problems which are typically NP-hard. In this talk, we’ll focus on a particular instance of this algorithm, called “Balanced Cuts,” by describing the cost function whose minimum coincides with the optimal partitioning, and by showing how a relaxation of the discrete minimization problem yields a continuous problem that can be solved through a generalization of iteratively reweighted least squares (IRLS).
Speaker Bio:
Nathan Cahill is an Associate Professor of Mathematical Sciences and the Director of the PhD Program in Mathematical Modeling at RIT. He earned BS and MS degrees in Applied Mathematics at RIT, attained the rank of Principal Scientist at Eastman Kodak and Carestream Health where he was granted 26 US patents, and earned a DPhil in Engineering Science from the University of Oxford. In 2009, he returned to his RIT roots and joined the faculty of the School of Mathematical Sciences, where he carries out research in overlapping areas between computational mathematics, mathematical modeling, computer science, and engineering. He is a SIAM Member, an IEEE Senior Member, and a SPIE Rising Researcher.
Intended Audience:
Undergraduates, graduates, and experts. Those with interest in the topic.
Keep up with DisCoMath Seminars on the DisCoMathS webpage.
Event Snapshot
When and Where
Who
This is an RIT Only Event
Interpreter Requested?
No