Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MD72

4 - Network Models for Multiobjective Discrete Optimization Merve Bodur, University of Toronto, 5 King’s College Rd., Toronto, ON, M5S 3G8, Canada, David Bergman, Andre Augusto Cire, Carlos Henrique Cardonha We propose a novel framework for solving multiobjective discrete optimization problems with an arbitrary number of objectives. Our framework formulates these problems as network models, in that enumerating the Pareto frontier amounts to solving a multicriteria shortest path problem in an auxiliary network. We design tools and techniques for exploiting the network model in order to accelerate the identification of the frontier. We show that the proposed framework yields orders-of-magnitude performance improvements over existing state-of-the-art algorithms on four problem classes containing both linear and nonlinear objective functions. n MD72 West Bldg 211A Joint Session Wagner/Practice: Daniel H. Wagner Prize for Excellence in Operations Research Practice I Emerging Topic: Daniel H. Wagner Competition Emerging Topic Session Chair: Patricia Neri, SAS Institute, Inc., 104 Grandtree Ct., Cary, NC, 27519, United States 1 - Combinatorial Exchanges for Trading Fishery Access Rights Martin Bichler, Technische Universitat Munchen, Department of Informatics, Boltzmannstr. 3, Munich, 85748, Germany, Douglas Ferrell, Jacob K. Goeree Overfishing is a prime environmental concern. Catch share systems have recently been shown to be effective tools to combat overfishing. Yet, the allocation of catch shares has always been a challenging policy problem. There is an active discussion about market-based solutions for the allocation and re-allocation of fishery shares. Unfortunately, until yet there have not been adequate market designs to address the specific requirements in these markets. The recent subsidized share trading market in New South Wales (NSW) is a first-of-a- kind market design for the reallocation of catch shares and the largest combinatorial exchange to date. The market design needed to address several non-standard requirements, most importantly the lack of participation and fair payments. While these features were crucial for the adoption of the proposed design, they led to computationally challenging allocation and pricing problems. The implemented exchange illustrates how computational optimization and market design can provide new policy tools, able to solve complex policy problems considered intractable only a few years ago. The exchange operated from May to July 2017 and effectively reallocated shares from inactive fishers to those who needed them most. It can provide a template for the reallocation of catch shares in other fisheries world- wide as well. 2 - Collaborative Human-UAV Search and Rescue for Missing Tourists in Nature Reserves Yu-Jun Zheng, Hangzhou Normal University, Hangzhou, China, Wei-Guo Sheng, Yi-Chen Du, Hai-Feng Ling The use of unmanned aerial vehicles (UAVs) is becoming commonplace in search and rescue tasks in complex terrains. In the literature, there are a number of studies on UAV search with the objective of minimizing search time and/or maximizing detection probability. However, little effort has been devoted to collaborative human and UAV search, which is necessary in many applications where the target has to be ultimately reached by human rescuers. In this paper we present a collaborative human-UAV search planning problem with the aim of minimizing the expected time at which the target is reached by human rescuers. The presented problem is of high complexity, and thus traditional exact algorithms would be very time-consuming or even impractical for solving even relatively small instances. We propose an evolutionary algorithm which uses biogeography-inspired operators to efficiently evolve a population of solutions to find the optimum or a near-optimum within an acceptable time. Computational experiments demonstrate the advantages of our algorithm over a number of other popular algorithms. The proposed method has been successfully applied to two real-world operations for searching and rescuing missing tourists in a nature reserve in China. It is estimated that, compared to the old method used by the organization, our method shortened the time required for reaching the targets by 79 minutes and 147 minutes in the two cases, respectively, providing a great improvement in the life-critical operations.

n MD73 West Bldg 211B JFIG Panel Discussion: Tips for Writing CAREER Proposals Sponsored: Junior Faculty JFIG Sponsored Session Chair: Ehsan Salari, Wichita State University, Wichita, KS, 67260, United States Co-Chair: Anahita Khojandi, University of Tennessee, Knoxville, TN, 37996, United States 1 - JFIG Panel Discussion: Tips for Writing CAREER Proposals Ehsan Salari, Wichita State University, 120C Engineering Building, 1845 Fairmount St., Wichita, KS, 67260, United States Recent NSF CAREER Awardees will share their experience submitting their award winning proposals, give advice and answer questions. Panelists Jennifer A. Pazour, Rensselaer Polytechnic Institute, 110 8th street, CII 5217, Troy, NY, 12180, United States Amin Khademi, Clemson University, Central, SC, 29630, United States Van-Anh Truong, Columbia University, 500 West 120th Street, 338 Mudd Hall, New York, NY, 10027, United States Andy Sun, Georgia Institute of Technology, 755 Ferst Drive, Atlanta, GA, 30312, United States n MD74 West Bldg 212A Searching the Solution Space in Multiple Criteria Optimization Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Banu Lokman, Middle East Technical University, Ankara, 06800, Turkey Co-Chair: Murat Mustafa Koksalan, Middle East Technical University, Ankara, 06531, Turkey 1 - A Multi Objective Approach to Clustering Data with Heterogeneous Inputs Banu Lokman, Middle East Technical University, Department of Industrial Engineering, Universiteler Mahallesi, Ankara, 06800, Turkey, Dilay Aktas, Tulin Inkaya Clustering algorithms mostly use a single dissimilarity matrix to partition a set of objects into a set of groups such that the objects assigned to the same group are similar for some criteria. When more than one dissimilarity matrix is available, many applications aggregate the matrices to come up with a single matrix, possibly masking the true nature of the original data. In this study, we develop a multi-objective approach to generate a number of clustering solutions considering different dissimilarity matrices. 2 - Representing the Nondominated Set for Multi-objective Integer Programs Ilgin Dogan, University of California Berkeley, Berkeley, CA, 94708, United States, Banu Lokman, Murat Mustafa Koksalan Multi-Objective Integer Programs have a wide variety of application areas. Since the number of nondominated points grows exponentially with the problem size and finding all points is hard, generating a subset with “desired properties is important. In this study, we observe that the distribution of nondominated points is critical in defining the desired properties of the representative subset. Based on our observations, we develop algorithms to generate a subset of points that represents the nondominated set with a prespecified coverage error. Our computational experiments show that our algorithms outperform the existing ones in terms of cardinality of the representative set and solution time.

240