Remote Sensing and Earth Observation
Onboard AI for early fire-smoke detection using hyperspectral satellite imagery, with attention to lightweight and energy-efficient deployment on constrained satellite platforms.
Applied AI | Data Analytics | Earth Observation
Dr Sha Lu
Research Associate, School of Computer Science and Information Technology, Adelaide University
I develop machine learning methods and applied AI systems for remote sensing, biomedical signal analysis, and anomaly detection, with current work spanning onboard satellite fire-smoke detection and seizure prediction from EEG data.
82
Google Scholar citations
5
h-index
3
i10-index
26+
International patents from industry work
Profile
Research shaped by methodological rigor and real-world deployment.
Dr Sha Lu is a Research Associate in data analytics and applied AI. Her work connects machine learning methods with high-impact applications in satellite Earth observation, biomedical signal analysis, and anomaly detection.
She received her PhD in Data Science from the University of South Australia in 2021 and brings more than 20 years of combined academic and industry experience, including software engineering and project management at a leading IT company.
Research
Three connected research themes.
Biomedical Applications
Predictive modelling of epileptic seizures using long-term intracranial EEG and scalp EEG data, including deep learning, signal processing, channel coherence, and interpretable time-series modelling.
Anomaly Detection
Dependency-, proximity-, and probabilistic modelling for rare and abnormal event detection, including LogDP, LoPAD, and broader dependency-based anomaly detection frameworks.
Projects
Current and recent funded research.
2024-2026
On-orbit fire-smoke detection on Kanyini and Phi-Sat-2
On-orbit evaluation and demonstration of energy-efficient fire smoke detection using HS2 imagery and onboard AI. Funded by SmartSat CRC.
2023-2024
Epileptic seizure prediction with long-term iEEG
Deep learning models for seizure prediction using long-term intracranial EEG recordings. Supported by the ARC Training Centre in Cognitive Computing for Medical Technologies.
2022-2023
SmartSat P2-38 onboard AI for early fire-smoke detection
Energy-efficient onboard AI research for early detection of fire smoke from satellite imagery. Funded by SmartSat CRC.
Publications
Selected publications and recent outputs.
2026
Dependency-based anomaly detection: A general framework and comprehensive evaluation
Sha Lu, Lin Liu, Kui Yu, Thuc Duy Le, Jixue Liu, Jiuyong Li. Expert Systems with Applications, 297, 129249.
DOI2026
uLEAD-TabPFN: Uncertainty-aware Dependency-based Anomaly Detection with TabPFN
Sha Lu, Jixue Liu, Stefan Peters, Thuc Duy Le, Craig Xie, Lin Liu, Jiuyong Li. CoRR, abs/2604.20255.
arXiv2025
Leveraging Channel Coherence in Long-Term iEEG Data for Seizure Prediction
Sha Lu, Lin Liu, Jiuyong Li, Jordan D. Chambers, Mark J. Cook, David B. Grayden. IEEE Journal of Biomedical and Health Informatics, 29(8), 5541-5548.
DOI2025
Can EEG Foundation Models Help with Epileptic Seizure Prediction?
Xudong Guo, Lin Liu, Sha Lu, Jiuyong Li, Thuc Duy Le, Jixue Liu. IEEE Big Data, 1924-1933.
DOI2024
Onboard AI for Fire Smoke Detection Using Hyperspectral Imagery
Sha Lu, Eriita G. Jones, Liang Zhao, Yu Sun, A. K. Qin, Jixue Liu, Jiuyong Li, Prabath Abeysekara, Norman Mueller, Simon Oliver, Jim O'Hehir, Stefan Peters. IEEE JSTARS, 17, 9629-9640.
DOI2021
Divide and Conquer: Targeted Adversary Detection using Proximity and Dependency
Sha Lu, Lin Liu, Jiuyong Li, Thuc Duy Le, Jixue Liu. ICBK, 125-132.
DOI2020
LoPAD: A Local Prediction Approach to Anomaly Detection
Sha Lu, Lin Liu, Jiuyong Li, Thuc Duy Le, Jixue Liu. PAKDD, 660-673.
DOI2018
Effective Outlier Detection based on Bayesian Network and Proximity
Sha Lu, Lin Liu, Jiuyong Li, Thuc Duy Le. IEEE Big Data, 134-139.
DOIResearch Indicators
Scholar metrics captured on 2026-06-16.
Google Scholar showed 82 total citations, 79 citations since 2021, h-index 5, and i10-index 3 at the time this site content was prepared.
Current Google Scholar profileTeaching and Supervision
Analytics teaching and HDR co-supervision.
Teaching
- Lecturer, INFS 5102 Unsupervised Methods in Analytics, UniSA, 2022.
- Practical Supervisor Tutor, INFS 5102 Unsupervised Methods in Analytics, UniSA, 2018-2022.
Supervision
Eligible to supervise Masters and PhD.
- Transformer-based causal inference methods for temporal data with latent confounders.
- Remote Sensing Foundation Models for Wildfire Smoke Detection.
Professional Service
Service to the research community.
2026
Program Committee member
IEEE International Conference on Data Mining (ICDM 2026).
Background
Education and professional experience.
PhD in Data Science
University of South Australia, STEM.
Software engineering and project management
Leading IT company, contributing to more than 26 international patents in wireless communication.
Master, Pattern Recognition
Sichuan University.
Bachelor, Computer Science
Sichuan University.
Contact