Faculty Research Award Recipients
2003: Bryant Julstrom
Dr. Bryant Julstrom, Professor of Computer Science (Photo taken by Neil Andersen, University Photographer)
The 2002-2003 College of Science and Engineering (COSE) Faculty Research Award was presented to Dr. Bryant Julstrom, Professor of Computer Science, on August 27, 2003. The COSE Faculty Research Award was established in 1997-1998 and it recognizes outstanding contributions to research during a faculty member’s appointment within COSE. Award criteria include a faculty member’s collaboration with and support of student researchers, dissemination of research results (publications and presentations), external recognition of the quality of research activities and service to the professional community.
Julstrom’s research focuses on evolutionary computation, in particular the application of genetic algorithms to combinatorial problems. An algorithm is a procedure or series of steps that can be used to solve a problem; in Computer Science, an algorithm describes the logical sequence of operations to be performed by a program. Julstrom states that, “An evolutionary algorithm (EA) is a probabilistic search algorithm inspired by the essential features of biological evolution: reproduction with variation, selection according to fitness, and repetition. An EA maintains a population of data structures, called chromosomes, that represent candidate solutions to its target problem. It [the EA] applies operators based on genetic crossover and mutation to generate new data structures, representing new candidate solutions, and as generations of these chromosomes succeed each other, better solutions evolve.”
Some problems are computationally easy and the algorithms that solve these problems exactly require times that grow only as polynomial functions of the sizes of problem instances. “However,” observes Julstrom, “many interesting and useful problems are computationally hard. All known exact algorithms for these problems require times that grow exponentially with problem size, so for large instances, exact algorithms are infeasible. That is, we know how to get a solution to a large instance of such a problem, but it would take far too long for even the fastest computer to finish the necessary computation.” For problems such as these, “we turn to heuristics, algorithms that return approximate solutions reasonably quickly.”
Over the last 14 years Julstrom has collaborated with four undergraduate and three graduate students on seven publications; supervised six Master of Science theses and starred papers; reviewed submissions to journals and conferences; and published over 60 papers. Most of the papers address aspects of evolutionary computation, particularly applications to problems of combinatorial optimization. At the 2001 Genetic and Evolutionary Computation Conference workshop on Representations and Operators for Network Problems his paper “The Blob Code: A Better String Coding of Spanning Trees for Evolutionary Search” was named the best paper.
Julstrom was the Chair of the Evolutionary Computation and Optimization track of the 2004 Association for Computing Machinery (ACM) Symposium on Applied Computing which was held in Nicosia, Cyprus, in March, 2004. Julstrom and Athos Antoniades co-authored a paper titled "Two Hybrid Evolutionary Algorithms for the Rectilinear Steiner Arborescence Problem" that was accepted for presentation at the conference. Antoniades graduated Cum Laude in May, 2003, with a Bachelor of Science degree in Computer Science. He is currently enrolled at the University of Cyprus where he is pursuing a Ph.D. in Computer Science with an emphasis in Bioinformatics. As part of the research for their paper, Antoniades ran many of the tests of the algorithms. He also presented their paper at the conference.
Per Julstrom, the Rectilinear Steiner Arborescence (RSA) problem may be explained in this way, “In a rectilinear structure, all line segments are horizontal or vertical. Given a collection of points in the first quadrant of the plane, a RSA is a rectilinear tree that connects the points to the origin and in which every path from the origin leads only up and to the right. The minimum RSA problem seeks such a tree of minimum total length.” The RSA problem has applications in designing integrated circuits, where terminals must be connected to a source as economically as possible. The hybrid EAs that Julstrom and Antoniades developed “differ in how they represent the candidate arborescences and how they perform crossover and mutation.” In comparisons with a non-evolutionary heuristic, one of the EAs performed better, the other worse.
Antoniades states that his research experience with Julstrom provided him “the opportunity to learn from the way he works.” In addition, Antoniades offers that, “The quality of his work is excellent and having observed and taken part in this work gave me significant direction to help me become a better researcher. … He is an excellent educator and researcher.”