Current Search:   Boginski, Vladimir (x)

View All Items

  • CSV Spreadsheet
(1 - 3 of 3)
Methods for online feature selection for classification problems
Title
Methods for online feature selection for classification problems.
Creator
Razmjoo, Alaleh, Zheng, Qipeng, Rabelo, Luis, Boginski, Vladimir, Xanthopoulos, Petros, University of Central Florida
Abstract / Description
Online learning is a growing branch of machine learning which allows all traditional data miningtechniques to be applied on an online stream of data in real-time. In this dissertation, we presentthree efficient algorithms for feature ranking in online classification problems. Each of the methodsare tailored to work well with different types of classification tasks and have different advantages.The reason for this variety of algorithms is that like other machine learning solutions, there is...
Show moreOnline learning is a growing branch of machine learning which allows all traditional data miningtechniques to be applied on an online stream of data in real-time. In this dissertation, we presentthree efficient algorithms for feature ranking in online classification problems. Each of the methodsare tailored to work well with different types of classification tasks and have different advantages.The reason for this variety of algorithms is that like other machine learning solutions, there is usuallyno algorithm which works well for all types of tasks. The first method, is an online sensitivitybased feature ranking (SFR) which is updated incrementally, and is designed for classificationtasks with continuous features. We take advantage of the concept of global sensitivity and rankfeatures based on their impact on the outcome of the classification model. In the feature selectionpart, we use a two-stage filtering method in order to first eliminate highly correlated and redundantfeatures and then eliminate irrelevant features in the second stage. One important advantage of ouralgorithm is its generality, which means the method works for correlated feature spaces withoutpreprocessing. It can be implemented along with any single-pass online classification method withseparating hyperplane such as SVMs. In the second method, with help of probability theory wepropose an algorithm which measures the importance of the features by observing the changes inlabel prediction in case of feature substitution. A non-parametric version of the proposed methodis presented to eliminate the distribution type assumptions. These methods are application to alldata types including mixed feature spaces. At last, we present a class-based feature importanceranking method which evaluates the importance of each feature for each class, these sub-rankingsare further exploited to train an ensemble of classifiers. The proposed methods will be thoroughlytested using benchmark datasets and the results will be discussed in the last chapter.
Show less
Date Issued
2018
Identifier
CFE0007584, ucf:52567
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0007584
Multi-level Optimization and Applications with Non-Traditional Game Theory
Title
Multi-level Optimization and Applications with Non-Traditional Game Theory.
Creator
Yun, Guanxiang, Zheng, Qipeng, Boginski, Vladimir, Karwowski, Waldemar, Yong, Jiongmin, University of Central Florida
Abstract / Description
We study multi-level optimization problem on energy system, transportation system and information network. We use the concept of boundedly rational user equilibrium (BRUE) to predict the behaviour of users in systems. By using multi-level optimization method with BRUE, we can help to operate the system work in a more efficient way. Based on the introducing of model with BRUE constraints, it will lead to the uncertainty to the optimization model. We generate the robust optimization as the...
Show moreWe study multi-level optimization problem on energy system, transportation system and information network. We use the concept of boundedly rational user equilibrium (BRUE) to predict the behaviour of users in systems. By using multi-level optimization method with BRUE, we can help to operate the system work in a more efficient way. Based on the introducing of model with BRUE constraints, it will lead to the uncertainty to the optimization model. We generate the robust optimization as the multi-level optimization model to consider for the pessimistic condition with uncertainty. This dissertation mainly includes four projects. Three of them use the pricing strategy as the first level optimization decision variable. In general, our models' first level's decision variables are the measures that we can control, but the second level's decision variables are users behaviours that can only be restricted within BRUE with uncertainty.
Show less
Date Issued
2019
Identifier
CFE0007881, ucf:52758
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0007881
STOCHASTIC OPTIMIZATION AND APPLICATIONS WITH ENDOGENOUS UNCERTAINTIES VIA DISCRETE CHOICE MODELSl
Title
STOCHASTIC OPTIMIZATION AND APPLICATIONS WITH ENDOGENOUS UNCERTAINTIES VIA DISCRETE CHOICE MODELSl.
Creator
Chen, Mengnan, Zheng, Qipeng, Boginski, Vladimir, Vela, Adan, Yayla Kullu, Muge, University of Central Florida
Abstract / Description
Stochastic optimization is an optimization method that solves stochastic problems for minimizing or maximizing an objective function when there is randomness in the optimization process. In this dissertation, various stochastic optimization problems from the areas of Manufacturing, Health care, and Information Cascade are investigated in networks systems. These stochastic optimization problems aim to make plan for using existing resources to improve production efficiency, customer...
Show moreStochastic optimization is an optimization method that solves stochastic problems for minimizing or maximizing an objective function when there is randomness in the optimization process. In this dissertation, various stochastic optimization problems from the areas of Manufacturing, Health care, and Information Cascade are investigated in networks systems. These stochastic optimization problems aim to make plan for using existing resources to improve production efficiency, customer satisfaction, and information influence within limitation. Since the strategies are made for future planning, there are environmental uncertainties in the network systems. Sometimes, the environment may be changed due to the action of the decision maker. To handle this decision-dependent situation, the discrete choice model is applied to estimate the dynamic environment in the stochastic programming model. In the manufacturing project, production planning of lot allocation is performed to maximize the expected output within a limited time horizon. In the health care project, physician is allocated to different local clinics to maximize the patient utilization. In the information cascade project, seed selection of the source user helps the information holder to diffuse the message to target users using the independent cascade model to reach influence maximization. \parThe computation complexities of the three projects mentioned above grow exponentially by the network size. To solve the stochastic optimization problems of large-scale networks within a reasonable time, several problem-specific algorithms are designed for each project. In the manufacturing project, the sampling average approximation method is applied to reduce the scenario size. In the health care project, both the guided local search with gradient ascent and large neighborhood search with Tabu search are developed to approach the optimal solution. In the information cascade project, the myopic policy is used to separate stochastic programming by discrete time, and the Markov decision process is implemented in policy evaluation and updating.
Show less
Date Issued
2019
Identifier
CFE0007792, ucf:52347
Format
Document (PDF)
PURL
http://purl.flvc.org/ucf/fd/CFE0007792