Abstract: Classification of underwater objects such as unexploded ordnances (UXO) and mines from sonar datasets poses a difficult problem. Among factors that complicate classification of these objects are: variations in the operating and environmental conditions, presence of spatially varying clutter, variations in target shape, composition, orientation and burial conditions. Furthermore, collection of large quantities of real and representative data for training and testing in various background conditions is very difficult and impractical in many cases. In this dissertation, we build on our previous work in my MS Thesis where sparse-reconstruction based classification models were trained on synthetically generated sonar datasets to perform classification on real datasets. While this earlier work helped address issues of data poverty that are intrinsic to the underwater mine-hunting problem, in this work we change course to focus on the adaptation of such models. Particularly, we investigate approaches to adapting linear and kernelized forms of sparse reconstruction based classifiers (SRCs) to function in a *lifelong learning* setting, in order to perform classification as environmental parameters are constantly evolving, without sacrificing performance on previously encountered environments.

Our solution combines Incremental Kernel Embeddings, Ridge Leverage Score (RLS) Sampling and Incremental Sparse Reconstruction Classifiers (SRC) to efficiently adapt both our features and our decision models over the system's operational life. Results for sampling and classification on sonar datasets are discussed and analyzed.

Adviser: Mahmood R. Azimi-Sadjadi
Co-Adviser: n/a
Non-ECE Member: Michael Kirby
Member 3: Ali Pezeshki
Addional Members: J. Rockey Luo

Publications:
"Adaptive Classification Using Incremental Linearized Kernel Embedding", John J. Hall, C. Robbiano, and M. R. Azimi-Sadjadi. IEEE Transactions on Signal Processing (In Review).

"Incremental Dictionary Learning For Adaptive Classification And Reconstruction Of Facial Imagery."
M. R. Azimi, C. Robbiano, Y. Zhao, J. Hall Proceedings of IEEE MLSP ’19 Pittsburgh, PA. October 2019.

"Analyzing Transfer Learning Methods For UXO Classification In Varying Shallow Water Environments." N. Larson, J. Hall, M. R. Azimi Proceedings of IEEE MLSP ’19 Pittsburgh, PA. October 2019.

"Underwater UXO Classification using A Matched Subspace Classifier with Adaptive Dictionaries."
Jack Hall, Mahmood R. Azimi-Sadjadi, et. al, IEEE Transactions on Oceanic Engineering June 2018.

"Automatic Equatorial GPS Amplitude Scintillation Detection Using a Machine Learning Algorithm."
Yu Jiao, John J. Hall, and Yu T. Morton. IEEE Transactions on Aerospace and Electronic Systems April 2017.

Program of Study:
ECE 511
STAT 525
ECE 520
MATH 676
STAT 620
ECE 795
ECE 799 (26 credits)
N/A