Name of Visiting Scholar: Nesrine Amor
Current Position/Title: Senior Researcher/Dr.
Institutional Affiliation: Technical University of Liberec, Czech Republic
Email: nesrine.amor@tul.cz
Webpage: https://www.linkedin.com/in/nesrine-amor-478b3a86/
Host Scholar: Kuang-Yow Lian, Distinguished Professor
Hosting Department/Institution: Department of Electrical Engineering, National Taipei University of Technology.
Biography:
Dr. Nesrine Amor is a Senior Researcher at the Technical University of Liberec, Czech Republic. Her research combines electrical engineering, statistical signal processing, artificial intelligence, and computational intelligence to address complex engineering challenges. Her expertise spans Bayesian state estimation, machine learning, intelligent optimization, statistical modeling, and data-driven engineering, with applications in biomedical signal processing, power systems, advanced materials, and manufacturing. In recent years, her research has focused on integrating machine learning, intelligent optimization, and generative artificial intelligence to accelerate the modeling, design, and optimization of advanced materials and engineering systems. Dr. Amor has authored more than 40 peer-reviewed publications in leading international journals and conferences and has an H-index of 21. She has contributed to several international and EU-funded research projects in signal processing, artificial intelligence, autonomous systems, and advanced materials, and actively serves as a reviewer and guest editor for high-impact scientific journals. Her current research focuses on developing intelligent computational frameworks that integrate Bayesian inference, machine learning, and optimization to solve challenging problems across electrical engineering and modern engineering systems.
Lecture [1]:
Date ＆ Time: August, 5, 2026 at 14.30-16.00
Venue: Conference room, Department of Electrical Engineering, National Taipei University of Technology.
Title: Bayesian State Estimation for Intelligent Engineering Systems
Abstract:
Many engineering systems are governed by dynamic processes whose internal states cannot be measured directly. Instead, these hidden states must be inferred from noisy, incomplete, and often nonlinear measurements acquired by sensors. From power systems and biomedical engineering to robotics and autonomous technologies, accurate state estimation is essential for understanding the behavior of complex systems and extracting reliable information from measured data.
This lecture presents Bayesian state estimation as a unified probabilistic framework for estimating hidden states in nonlinear dynamic systems. Beginning with the principles of Bayesian inference and state-space modeling, it reviews the evolution of modern state estimation techniques, from the Kalman Filter and Unscented Kalman Filter to the Particle Filter for nonlinear and non-Gaussian systems. Particular emphasis is placed on constrained Bayesian estimation, demonstrating how prior knowledge and physical constraints can be incorporated into the estimation process to improve robustness while ensuring mathematically consistent and physically meaningful solutions.
The lecture combines theoretical developments with representative engineering applications, including dynamic state estimation in power systems using phasor measurement unit (PMU) data, EMG-based movement recognition for prosthetic control, EEG source localization, and other nonlinear estimation problems arising from sensor measurements. These examples demonstrate how diverse engineering problems can be formulated within the same Bayesian framework, providing robust solutions for nonlinear, non-Gaussian, and constrained estimation problems.
Although the presented applications originate from different engineering domains, they share the same underlying estimation challenge: inferring hidden dynamic states from uncertain observations. This lecture demonstrates how Bayesian state estimation provides a unified mathematical framework for addressing these challenges across a wide range of engineering systems.
Lecture [2]:
Date ＆ Time: September, 9, 2026 at 14.00-15.30
Venue: International Conference Hall, Department of Electrical Engineering, National Taipei University of Technology.
Title: Machine Learning and Intelligent Optimization for Data-Driven Engineering Systems
Abstract:
Artificial intelligence is transforming engineering by changing how complex systems are modeled, optimized, and used to support decisions. Engineers now have access to growing volumes of data from experiments, sensors, and simulations, but converting these data into reliable engineering knowledge remains a major challenge. Machine learning can capture nonlinear relationships and predict system responses, while intelligent optimization can use these predictions to search for improved designs and operating conditions.
This lecture presents an integrated framework for data-driven engineering, beginning with data preparation, feature representation, and artificial neural network modeling. It then addresses model validation, generalization, and performance on unseen data before introducing surrogate-assisted optimization, metaheuristic search, engineering constraints, and multi-objective decision-making. Particular attention is given to balancing exploration and exploitation and to experimentally validating optimized solutions.
The framework is illustrated through case studies in advanced materials and manufacturing. These include predicting the functional properties of nano-titanium-dioxide-coated cotton, optimizing a neural network with the Golden Eagle Optimizer to predict the comfort properties of zinc-oxide-coated fabrics, and combining Grey Relational Analysis with the White Shark Optimizer to improve machining conditions for glass-fiber-reinforced polymer composites.
Although the applications focus on advanced materials and manufacturing, the underlying methodology is broadly applicable across engineering domains. The central message is straightforward: machine learning predicts what may happen, optimization identifies promising choices, and engineering validation confirms whether those choices work in practice.