Heidar A. Malki is currently a Professor of Engineering Technology Department and Senior Associate Dean of the Technology Division at the Cullen College of Engineering. He also has a joint appointment with Electrical and Computer Engineering Department at UH. He holds a Ph.D. degree in Electrical Engineering from the University of Wisconsin-Milwaukee. He is a senior member of IEEE and associate editor for the IEEE Transactions on Fuzzy Systems. Dr. Malki was the general chair for the 1997 ASEE/GSW Conference and one of co-chairs of 1997 ICNN-IEEE International Conference on Neural Networks.
Title: Lithium-Ion Battery Prognostic and Health Prediction Using Machine Learning Models
Abstract: Lithium-Ion battery prognostic and health prediction is an essential part of our modern world today. Reliable predictions of the current state and remaining useful life are critical to ensure safe operation of various industrial applications. There exists established and extensively researched battery state prediction models including the equivalent circuit and physics-based models. The emergence and advancements of machine learning algorithms, such as support vector machine (SVM), have led to a new era of data-driven models. In this research, we initially establish a baseline prediction using traditional machine learning algorithms and subsequently employ deep learning algorithms. Our analysis focuses on the battery’s degradation pattern, specifically the predication of the State of Health (SOH) prediction and the Remaining Useful Life (RUL), aiming to predict the degradation from a specific threshold cycle to the EOL of the battery. In this study, our aim is to establish a comparable baseline for several machine learning-based prognostic predictions. To overcome this problem, we conduct several experiments on the same dataset using traditional algorithms and as well as neural network models. The classification accuracy aligns with several benchmarks established in the research literature. This research further underscores the viability of using Long Short-Term Memory (LSTM) models for RUL prediction, thereby advancing the role of data-driven models in prognostic and health management for critical engineering applications.
Title: SecureReg: A Combined Framework for Proactively Exposing Malicious Domain Name Registrations
Abstract: —Rising cyber threats, with miscreants registering thousands of new domains daily for Internet-scale attacks like spam, phishing, and drive-by downloads, emphasize the need for innovative detection methods. This paper introduces a cuttingedge approach for identifying suspicious domains at the onset of the registration process. The accompanying data pipeline generates crucial features by comparing new domains to registered domains, emphasizing the crucial similarity score. Leveraging a novel combination of Natural Language Processing (NLP) techniques, including a pretrained CANINE model, and Multilayer Perceptron (MLP) models, our system analyzes semantic and numerical attributes, providing a robust solution for early threat detection. This integrated Pretrained NLP (CANINE) + MLP model showcases outstanding performance, surpassing both individual pretrained NLP models and standalone MLP models. With an impressive F1 score of 84.86% and an accuracy of 84.95% on the SecureReg dataset, it effectively detects malicious domain registrations. The findings demonstrate the effectiveness of the integrated approach and contribute to the ongoing efforts to develop proactive strategies to mitigate the risks associated with illicit online activities through the early identification of suspicious domain registrations.