07/06/2022
By Parisa Hajibabaee

The Kennedy College of Sciences, Department of Computer Science, invites you to attend a Doctoral Dissertation Proposal defense by Parisa Hajibabaee on “Machine Learning Applications in Structural Engineering.”

Ph.D. Candidate: Parisa Hajibabaee
Defense Date: Friday, July 22, 2022
Time: 1 to 2:30 p.m. EST
Location: This will be a virtual defense via Zoom. Those interested in attending should contact parisa_hajibabaee@student.uml.edu and committee advisor, farhad_pourkamali@uml.edu, at least 24 hours prior to the defense to request access to the meeting.

Committee:
Committee Chair (Advisor): Farhad Pourkamali-Anaraki, Ph.D., Assistant Professor, Computer Science, UMass Lowell
Reza Ahmadzadeh, Ph.D., Assistant Professor, Computer Science, UMass Lowell
Scott Stapleton, Ph.D., Associate Professor, Mechanical Engineering, UMass Lowell
MohammadAmin Hariri-Ardebili, Ph.D., Affiliated Research Associate, Engineering and Applied Science, University of Colorado Boulder

Abstract:
The recent successful machine learning applications in many scientific research areas such as healthcare, advertising and agriculture have sparked a domino effect, prompting others to wonder whether their fields of practice, including civil engineering applications, may be altered or revolutionized by machine learning. For example, structural engineering researchers have been actively looking into machine learning to solve complex engineering problems over the last few years. The main objective of this thesis is to apply machine learning methods that have received significant attention to the structural engineering discipline. We mainly concentrate on the intersection of machine learning and structural engineering applications within three projects:

The first project focuses on the problem of producing accurate and efficient low-rank approximations of kernel matrices that appear in a wide range of nonlinear machine learning techniques. We propose a new landmark selection technique called Distance-based Importance Sampling and Clustering (DISC), in which the relative importance scores are computed for improving accuracy-efficiency tradeoffs compared to existing works that range from probabilistic sampling to clustering methods. The proposed landmark selection method follows a coarse-to-fine strategy to capture the intrinsic structure of complex data sets, allowing us to substantially reduce the computational complexity and memory footprint with minimal loss in accuracy.

The second project looks at the performance of various dimensionality reduction techniques for analyzing and visualizing high-dimensional data sets generated from a computer simulation in structural engineering that characterizes risks posed by earthquakes to structures. We also study the effect of class-imbalanced data on the resultant low-dimensional embedding supplied by each dimensionality reduction method in a systematic manner.

The third project proposes to apply uncertainty quantification methods for machine learning models with application to structural engineering. The concept of uncertainty in machine learning models has received significant attention in the last two decades. One solution to quantify the uncertainty of prediction is to provide the prediction intervals rather than point estimation. We plan to use jackknife method and its variants, which are based on the construction of a set of leave-one-out models, to estimate robust prediction intervals with any kind of machine learning model.

All interested students and faculty members are invited to attend the defense via remote online access.