University calendar

ELEE Research Component of PhD Qualifier Exam By TahminaTabassum Treena

Thursday, December 10, 2026 at 2:30pm to 3:30pm

Lester W. Cory Conference Room, Science & Engineering Building (SENG), Room 213A : Zoom Link: https://umassd.zoom.us/j/95462387387 Meeting ID: 954 6238 7387 Passcode: 719961
Mohammad Karim
508.999.8434
mkarim@umassd.edu
https://umassd.zoom.us/j/95462387387

Topic: Geometry-Aware Machine Learning for Photonic Device Modeling and Optimization

Abstract: The design and optimization of photonic devices are challenging due to complex electromagnetic interactions and the strong dependence of optical responses on device geometry, material properties, and operating conditions. Although numerical simulations provide valuable physical insight, repeated evaluations across multidimensional design spaces are computationally expensive. Furthermore, conventional data-driven approaches often face difficulties in accurately predicting nonlinear resonance behavior for previously unseen device configurations. This research investigates a geometry-aware machine learning framework that integrates electromagnetic simulation, systematic parameter exploration, and robust surrogate modeling for photonic device design and optimization. The proposed methodology incorporates physically meaningful geometric features to improve the prediction of optical responses and examines multiple learning algorithms to evaluate predictive accuracy, robustness, and generalization. Particular emphasis is placed on resonance-focused learning, geometry-based validation, and the relationship between structural parameters and device performance. The framework is initially demonstrated using a plasmonic refractive-index sensing platform, where resonator dimensions, geometric asymmetry, and waveguide coupling influence resonance characteristics, electromagnetic field confinement, and sensing performance. Systematic and randomized numerical studies are combined with machine-learning-assisted design exploration, while electromagnetic field analysis, spectral refinement, and mesh comparisons are employed to assess physical behavior and numerical reliability. By integrating geometry-aware learning, resonance-focused prediction, and systematic electromagnetic validation, this research establishes a foundation for computationally efficient, physically grounded, and adaptable machine learning methodologies for photonic device modeling and optimization, with potential applications in optical sensing, integrated photonics, and advanced optical systems.

Advisor(s): Dr. Mohammad Karim, Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth

Committee Members: Dr. Yifei Li, Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth; Dr. Liudong Xing, Commonwealth Professor, Dept. of Electrical & Computer Engineering, UMASS Dartmouth; Dr. Tariq Manzur, Scientist & Engineer, Naval Undersea Warfare Center (NUWC) and Adjunct Professor, Department of Electrical & Computer Engineering, UMASS Dartmouth

NOTE: All ECE Graduate Students are ENCOURAGED to attend. All interested parties are invited to attend. Open to the public.

*For further information, please contact Dr. Mohammad Karim via email at mkarim@umassd.edu.

Back to top of page