Master of Science Project Defense: Pratik Kiran Bhansali
When: Wednesday,
December 14, 2016
8:30 AM
-
10:30 AM
Where: CCB 115
Cost: Free
Description: TOPIC: A PERFORMANCE ANALYSIS OF ALGORITHMS FOR THE PARETO
SOFTWARE RELIABILITY GROWTH MODEL
LOCATION: Charlton College of Business, CCB-115
ABSTRACT:
Software reliability engineering provides various models and techniques to estimate failures and ensure that software operates in a failure-free manner. It employs methods from probability theory and stochastic processes to model the reliability of a system with respect to a specific operational environment and input conditions. Many software reliability growth models (SRGM) are modeled as a Non-homogenous Poisson process (NHPP). Such NHPP SRGM can provide quantitative measures of the reliability of software systems. SRGM are used to estimate software reliability and predict future failures. They can also be used to track reliability improvement during software testing and correction.
To assess software reliability with a model, one must estimate the model parameters of NHPP based SRGM from failure data obtained during testing. Maximum Likelihood Estimation (MLE) is a procedure to estimate the parameters of a model from the data. There are various computational procedures to identify the MLEs of NHPP SRGM. These procedures include the expectation maximization (EM) algorithm, expectation conditional maximization (ECM) algorithm, and Newton’s method. The EM algorithm exhibits more stable convergence properties. However, the complexity of this algorithm increases as the number of model parameters increases, whereas Newton’s method is sensitive to the initial parameter estimates and can fail to converge if the initial estimates are not close to the MLEs. The ECM reduces the mathematical complexity of the EM algorithm by reducing the maximization (M)-step to multiple conditional maximization (CM)-steps. Hybrid algorithms use a combination of the ECM algorithm for a predefined number of iterations, followed by Newton’s method. This project compares the performance of these alternative algorithms in the context of the Pareto model for a variety of datasets from the research literature.
NOTE: All ECE Graduate Students are ENCOURAGED to attend.
All interested parties are invited to attend. Open to the public.
Advisor: Dr. Lance Fiondella
Committee Members: Dr. Liudong Xing and Dr. Paul J. Gendron, Department of Electrical & Computer Engineering
*For further information, please contact Dr. Lance Fiondella at 508.999.8596, or via email at lfiondella@umassd.edu.
SOFTWARE RELIABILITY GROWTH MODEL
LOCATION: Charlton College of Business, CCB-115
ABSTRACT:
Software reliability engineering provides various models and techniques to estimate failures and ensure that software operates in a failure-free manner. It employs methods from probability theory and stochastic processes to model the reliability of a system with respect to a specific operational environment and input conditions. Many software reliability growth models (SRGM) are modeled as a Non-homogenous Poisson process (NHPP). Such NHPP SRGM can provide quantitative measures of the reliability of software systems. SRGM are used to estimate software reliability and predict future failures. They can also be used to track reliability improvement during software testing and correction.
To assess software reliability with a model, one must estimate the model parameters of NHPP based SRGM from failure data obtained during testing. Maximum Likelihood Estimation (MLE) is a procedure to estimate the parameters of a model from the data. There are various computational procedures to identify the MLEs of NHPP SRGM. These procedures include the expectation maximization (EM) algorithm, expectation conditional maximization (ECM) algorithm, and Newton’s method. The EM algorithm exhibits more stable convergence properties. However, the complexity of this algorithm increases as the number of model parameters increases, whereas Newton’s method is sensitive to the initial parameter estimates and can fail to converge if the initial estimates are not close to the MLEs. The ECM reduces the mathematical complexity of the EM algorithm by reducing the maximization (M)-step to multiple conditional maximization (CM)-steps. Hybrid algorithms use a combination of the ECM algorithm for a predefined number of iterations, followed by Newton’s method. This project compares the performance of these alternative algorithms in the context of the Pareto model for a variety of datasets from the research literature.
NOTE: All ECE Graduate Students are ENCOURAGED to attend.
All interested parties are invited to attend. Open to the public.
Advisor: Dr. Lance Fiondella
Committee Members: Dr. Liudong Xing and Dr. Paul J. Gendron, Department of Electrical & Computer Engineering
*For further information, please contact Dr. Lance Fiondella at 508.999.8596, or via email at lfiondella@umassd.edu.
Contact:
ECE: Electrical & Computer Engineering Department 508.999.9164 http://www.umassd.edu/engineering/ece/
Topical Areas: General Public, University Community, College of Engineering, Electrical and Computer Engineering