2016 INFORMS Annual Meeting Program

418 / 561

INFORMS Nashville – 2016
418
2 - Justification And Justice Matter: Organizational Attention To Ideas
On Digital Innovation Platforms
Inchan Kim, University of New Hampshire, Durham, NH,
United States, John Qi Dong
Digital innovation platforms allow crowdsourcing of ideas, in effect generating big
data for organizations. Then, to which ideas do organizations direct their focus?
Based on the six logics of justification together with the four justice principles, we
use qualitative comparative analysis (QCA) to analyze the configurational impacts
of justification and justice reflected in ideas from on
organizational allocation of attention.
3 - Domain Specific Lexicon For Clinical Trial Subject
Eligibility Analysis
Euisung Jung, University of Toledo,
It is well understood that an NLP application requires sophisticated lexical
resources to support its processing goals. Different solutions have been proposed
to identify multi-gram disease named entities in the healthcare informatics
literature. In this study, we develop a domain-specific lexicon for n-gram Named
Entity Recognition (NER) in the breast cancer domain. The domain-specific
dictionary was evaluated by comparing it with Systematized Nomenclature of
Medicine—Clinical Terms (SNOMED CT). The results showed that it add
significant number of new terms which is very useful in effective natural
language processing.
WB66
Mockingbird 2- Omni
Graph Analytics for Complex Systems - II
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Hoang Tran, Texas A&M University, Address, College Station, TX,
United States,
Co-Chair: Satish Bukkapatnam, Texas A&M University, College Station,
College Station, TX, United States,
1 - Generalized Graph Shortest Path For Calibration Of
Computer Simulations.
Babak Farmanesh, Oklahoma State University,
Calibration refers to the process of adjusting parameters of a computer simulation
so that the simulation responses match the corresponding physical responses.
Calibration can be interpreted as a curve to surface matching problem. We
propose a graph-theoretic non-isometric matching approach to solve this problem
using the graph shortest path algorithm in one-dimensional spaces. For higher
dimensional spaces, we introduce the generalized shortest path concept to solve
the matching problem.
2 - Spectral Graph Theoretic Sensor Fusion
Prahalad Rao, Binghamton University,
The central theme of this talk is motivated from the compelling need for sensor-
based in situ quality assurance approaches in complex systems and processes,
such as additive manufacturing (AM). The key research question is: how to fuse
information from multidimensional sensor signals for monitoring and prognosis?
The proposed approach maps a multidimensional signal as an un-weighted
undirected network graph. Through this talk, it is demonstrated that graph
theoretic signal processing has the potential to monitor complex systems in a data
rich environment.
3 - Learning Data Association Graph
Chiwoo Park, Florida State University, 2525 Pottsdamer St.,
Tallahassee, FL, 32310, United States, ,
Taylor J. Woehl, James E. Evans, Nigel Browning
We presents a general formulation for a minimum cost data association problem
which associates data features via one-to-one, m-to-one and one-to-n links with
minimum total cost of the links. A motivating example is a problem of tracking
multiple interacting targets imaged on video frames. Many existing multitarget
tracking methods are capable of tracking non-interacting targets or tracking
interacting targets of restricted degrees of interactions. The proposed formulation
solves a multitarget tracking problem for general degrees of inter-object
interactions.
4 - Measuring Redundancy Of State Estimators In Large Networks By
Combining L1-minimziation And Integer Programming
Vishnu Vijayaraghavan, Texas A&M University, College Station,
TX, 77843, United States, , Kiavash Kianfar,
Yu Ding, Hamid Parsaei
Finding the degree of redundancy for structured linear systems is proven to be
NP-hard. Bound-and-decompose, 0-1 mixed integer programming (MIP) and
hybrid algorithms embedding 0-1 mixed integer programming within a bound-
and-decompose framework have all been studied and compared in the literature.
In this paper we take advantage of the computational efficiency of linear
programs to present an l1 minimization approach combined with mixed integer
programming to address this problem. This approach proves to be very effective in
measuring redundancy of state estimators in large networks.
WB67
Mockingbird 3- Omni
Joint Session QSR/DM: Process Monitoring for
Diverse Types of Data
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Youngseon Jeong, Annandale, VA, United States,
Co-Chair: Myong K Jeong, Rutgers University, Piscataway, NJ,
United States,
1 - Bayesian Based Distribution Free Procedure For Fault
Identification
Mehmet Turkoz, Rutgers University, 16 Rachel Terrace, Piscataway,
NJ, 08854, United States, Sangahn
Kim, Youngseon Jeong, Myong K Jeong, Elsayed A. Elsayed,
A.M.S. Hamouda, Khalifa Al-Khalifa
Many real life process control problems do not follow multivariate normal
distribution. In a process with unknown underlying distribution, identifying fault
variables of an out-of-control signal is a challenging issue for quality problems. In
this research, we present a new Bayesian fault identification method that does not
assume any specific probability distribution.
2 - Modeling And Shape Control Of Large Composite Components
For Section-to-fuselage Joints
Yuchen Wen, Gatech,
Shape control of large composite components is important in aerospace industry.
The current method of shape control has limitations of low efficiency and non-
optimal. We propose a surrogate-model based optimal shape control strategy in
order to achieve dimensional variation reduction and efficient shape adjustment
in large composite parts assembly process. The objective is accomplished by (i)
Investigating a surrogate model to achieve good prediction performance; (ii)
Conducting multi-objective optimization to determine the control actions from
the Pareto solutions; (iii) Implementing the sensitivity analysis to determine the
best number and positions of the locators.
3 - Process Tracking And Monitoring Based On Discrete
Jumping Model
Chao Wang, University of Wisconsin-Madison, Madison, WI,
United States, , Shiyu Zhou
Jumping model has been used as an effective tool in tracking and detecting
changes for continuous statistics in various applications. In this paper, we extend
the current jumping model from the continuous case to the discrete case to track
and monitor the changes in attribute data. The jumping model based posterior
distribution of the process mean is constructed with attribute data and prior
knowledge of the process. Using the posterior distribution, a control chart is then
developed to monitor the attribute data process.
4 - A New Bayesian Classification Model For Uncertain Data
Young-Seon Jeong, Chonnam National University, Gwangju,
Korea, Republic of, Byunghoon Kim,
Myong K Jeong, Jeongsub Choi, Soonmok Kwon, Jihoon Kang
This talk presents a new Bayesian classification model which considers the
correlation among uncertain features. Even though several classifiers for
uncertain data have been developed, they did not consider the dependency
among uncertain features, which have a critical effect on classification accuracy.
Experimental results with simulated data and real-life data show that the
proposed approach for uncertain data is more accurate than existing approaches.
WB66