2016 INFORMS Annual Meeting Program

385 / 561

INFORMS Nashville – 2016
385
WA64
Cumberland 6- Omni
Pareto Set Reduction Theories and Methods
Sponsored: Multiple Criteria Decision Making
Sponsored Session
Chair: Daniel Jornada, Texas A&M University, 1700 Research Parkway,
College Station, TX, 77843, United States,
1 - New Notions Of Efficiency For Multicriteria Optimization
Under Uncertainty
Devon Sigler, PhD Candidate, University of Colorado Denver,
Denver, CO, United States, ,
Alexander Engau
We present several new notions of efficiency for multicriteria optimization
problems under uncertainty. We demonstrate that these new definitions can be
fully characterized theoretically in a hierarchical manner and methodologically
using a collection of modified scalarization and generation methods. Related
computational comparisons and applications will be discussed.
2 - Robust Solutions To Uncertain Multiobjective Linear Programs
Garrett M. Dranichak, Department of Mathematical Sciences,
Clemson University, Clemson, SC, United States,
, Margaret M Wiecek
We study highly robust efficient solutions to multiobjective linear programs with
uncertainty in the objective function coefficients drawn from finite uncertainty
sets. We present results on existence and identification of highly robust efficient
solutions, as well as properties of and bounds on the highly robust efficient set.
Additional attention is given to a special case of problems yielding a robust
counterpart that is easily solvable.
3 - Pareto Set Reduction Theories And Methods
Kalyanmoy Deb, Department of Electrical and Computer
Engineering, Michigan State University,
In many multi-objective optimization problems, objectives become correlated to
each other thereby reducing the dimension of the Pareto-optimal set. In an
evolutionary multi-objective optimization method, we have integrated a principal
component analysis method to identify redundant objectives and solve very large-
scale problems.
4 - A Post-pareto Approach Using A Non-uniform Weight Generator
Method With Prioritized Objectives
Juan V Fernandez, Industrial, Manufacturing and Systems
Engineering Department, University of Texas at El Paso,
Multi-objective optimization has been recognized as an important research area in
the last years since many real life problems present multiple criteria that need to
be optimized simultaneously. Using metaheuristic methods or evolutionary
algorithms as solution methodologies leads to a large number of Pareto solutions
rather than a single unique optimum. Ultimately, all solutions are considered to
be Pareto-optimal and selecting the one solution among others can be an arduous
task for the decision-maker. This research presents a new developed approach
that uses a non-uniform weight generator method to reduce the size of the
Pareto-optimal set under the consideration of prioritized objectives.
WA65
Mockingbird 1- Omni
Digital Transformation of Labor, Media, Telecom, and
Financial Markets
Sponsored: Information Systems
Sponsored Session
Chair: Wei Chen, University of Arizona, University of Arizona,
Tucson, AZ, 85721, United States,
1 - What Do Employers Look For In Candidates?
Xuan Ye, New York University,
Prasanna Tambe
Using novel data with descriptions of job interview processes collected from a
career intelligence platform, I test the hypothesis that employers assess job
candidates’ ability, i.e. problem-solving skills and analytical skills, not labor
market experience when they recruit for jobs that require new technical skills.
These cognitive skills are hypothesized to be important in a fast moving
production environment. Employers’ evaluation methods are measured by text-
mining the interview questions contributed by the job candidates. With employer
fixed effects estimates, I find that employers use ability-based assessment question
39% more frequently for NewTech jobs than for other jobs.
2 - New Product Launch With Capacity Constraints And
Congestion-sensitive Consumers
Duy Dao, University of California, San Diego,
, Terrence August, Hyoduk Shin
A problem faced by the entertainment industry is the impact of congestion on the
release of a product. For theatrical releases, consumers have learned to delay
consumption, trading off this congestion cost with the loss of movie buzz as the
movie fades in relevance over time. Some may even forgo purchase because of
congestion. For online games, a surge of consumers logging on to play an
MMORPG can result in server issues during the initial release. In this paper, we
model how consumers decide when to make a purchase, considering the
congestion they experience. We then offer a strategy for how to profitably expand
the market when taking into consideration congestion-sensitive consumers.
3 - The Role Of Technological Discontinuity On
Incumbency Advantage
Xiahua Wei, University of Washington, Bothell,
This study examines how technological discontinuity contributes to market
competition, especially the incumbency of established firms. Based on the theory
of barrier to entry, we investigate the consequence of technological change in the
mobile telecommunications industry. We find that the ability of new entrants to
disrupt incumbents depends on the responsiveness of incumbents to the new
technologies. Further, we show that intense competition in the wake of
technological discontinuities, driven entirely by incumbents, can delay industry
shakeouts.
4 - Taxes And Equity Investment: Evidence From
Equity Crowdfunding
Wei Chen, University of Arizona, ,
Mingfeng Lin
Given the positive externality that entrepreneurial activities bring to the
economy, governments around the world have routinely resorted to various
incentives to spur entrepreneurship. In this paper, we empirically study whether
and how investments in early stage businesses respond to tax incentives, using a
natural experiment due to a policy change, and a comprehensive transaction-
level dataset from leading online equity crowdfunding platforms in the United
Kingdom.
WA66
Mockingbird 2- Omni
Data Analytics in Emerging Applications
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Nan Chen, National University of Singapore, Singapore,
Singapore,
1 - Inferring Three-dimensional Porous Defects Based On Cross-
sectional Images In Metal-based Additive Manufacturing
Jianguo Wu, Assistant Professor, University of Texas-El Paso, 500
W University Ave, Engineering Building, A-244, El Paso, TX,
79968, United States, , Nan Chen, Haijun Gong
Porosity is one of the most critical quality issues in the metal-based additive
manufacturing. This paper develops a novel quality inspection method by
inferring the size distribution, void density and volume fraction of 3D porosity
defects based on 2D cross-sectional images. The linkage between the size of
ellipsoidal defects and the size of cross-sectional elliptical contours is established.
An efficient Quasi-Monte Carlo EM algorithm is developed for 3D size
distribution estimation. The relationship between the 3D and 2D void densities is
developed to estimate the 3D void density and porosity. The effectiveness of the
proposed method is demonstrated through simulation and case studies.
2 - Change-point Detection On Solar Panel Performance Using
Thresholded Lasso
Youngjun Choe, University of Washington, Seattle, WA,
United States, , Weihong Guo, Eunshin Byon,
Judy Jin, Jingjing Li
Solar energy is a fast growing energy source. Solar energy stakeholders are,
however, concerned with sudden deterioration of photovoltaic systems’
performance. This study focuses on retrospectively identifying the time points of
abrupt changes. We present a nonparametric detection method based on
Thresholded LASSO. The proposed method is able to accurately detect
performance changes, while being robust against false detection under noisy
signals. The performance of the proposed method is evaluated and compared with
state-of-the-art methods through extensive simulations and a case study using
data collected from four solar energy facilities.
WA66