MASTER OF SCIENCE PROJECT DEFENSE BY: Varun Kumar Reddy Garlapati
When: Wednesday,
May 25, 2016
11:00 AM
-
1:00 PM
Where: Science & Engineering Building, Lester W. Cory Conference Room: Room 213A
Cost: free
Description: TOPIC: EXPECTATION CONDITIONAL MAXIMIZATION (ECM) ALGORITHMS FOR THE TRUNCATEDLOGISTIC AND TRUNCATED EXTREME VALUE MINIMUM SOFTWARERELIABILITY GROWTH MODELS
LOCATION: Lester W. Cory Conference Room, Science & Engineering Building (Group II), Room 213A
ABSTRACT:
Non-Homogeneous Poisson Process (NHPP)-based Software Reliability Growth Models (SRGM) can estimate different quantitative metrics such as the future number of failures and failure intensity. SRGM are complex parametric mathematical functions which are estimated from failure data with methods such as Maximum likelihood estimation (MLE) using different computational procedures such as Newton's Method. The Expectation Maximization (EM) algorithm enables MLE when the failure data is incomplete and possesses desirable properties such as monotonicity in convergence to the maximum likelihood estimate.
This project implemented Expectation Conditional Maximization (ECM) algorithms for the TLOGIS (Truncated Logistic Model) and TEVMIN (Truncated Extreme Value Minimum) SRGM. Initial parameter estimation techniques based on the EM algorithm were also developed. Convergence of the ECM was rapid, requiring 8-10 iterations. Robustness of the models to missing initial data was also considered. Deleting the initial 30 failures negatively impacted model prediction, but the models predicted future failures relatively well when 50 initial failures were removed.
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. Honggang Wang, Department of Electrical & Computer Engineering
*For further information, please contact Dr. Lance Fiondella at 508.999.8596, or via email at lfiondella@umassd.edu.
LOCATION: Lester W. Cory Conference Room, Science & Engineering Building (Group II), Room 213A
ABSTRACT:
Non-Homogeneous Poisson Process (NHPP)-based Software Reliability Growth Models (SRGM) can estimate different quantitative metrics such as the future number of failures and failure intensity. SRGM are complex parametric mathematical functions which are estimated from failure data with methods such as Maximum likelihood estimation (MLE) using different computational procedures such as Newton's Method. The Expectation Maximization (EM) algorithm enables MLE when the failure data is incomplete and possesses desirable properties such as monotonicity in convergence to the maximum likelihood estimate.
This project implemented Expectation Conditional Maximization (ECM) algorithms for the TLOGIS (Truncated Logistic Model) and TEVMIN (Truncated Extreme Value Minimum) SRGM. Initial parameter estimation techniques based on the EM algorithm were also developed. Convergence of the ECM was rapid, requiring 8-10 iterations. Robustness of the models to missing initial data was also considered. Deleting the initial 30 failures negatively impacted model prediction, but the models predicted future failures relatively well when 50 initial failures were removed.
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. Honggang Wang, 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