Ongoing research
Which constraints are beneficial?
Developing quality measures for clustering constraints and examining where existing approaches fall short.
Machine learning · Time series · Limited supervision
Hussein El Amouri
Machine Learning for Time Series: Methods, Interpretability, and Applications
Assistant Professor & AI Researcher
University of Sciences and Arts in Lebanon (USAL)
I study how expert knowledge and limited labels can make machine learning more useful and interpretable—from constrained clustering to seizure detection.
Edit this pageOpen to research collaborations, academic opportunities & industry R&D
01 / About
Research, teaching, and collaboration
I am an Assistant Professor and researcher at USAL in Lebanon. My research explores how limited supervision and expert knowledge can help us learn useful, interpretable representations of time-series data.
During my PhD at the University of Strasbourg, I developed Constrained DTW Preserving Shapelets (CDPS) and Shapelet Cluster Explanation (SCE). My current work asks which constraints are beneficial for clustering, and how their quality can be assessed reliably. I also collaborate with researchers at LAU on seizure detection using diffusion models and few-shot learning.
I welcome conversations about research collaborations, academic opportunities, and industry R&D.
02 / Research
Current and past research
Projects spanning time-series learning, constrained clustering, healthcare, speech synthesis, and supervised research.
Completed project
Community Aware Coded Caching
Supervision of a completed Lebanese University master’s project on community-aware coded caching.
Manuscript under revision
Seizure detection with limited labels
A collaboration with LAU exploring diffusion models and few-shot learning for epileptic seizure detection.
Published research
Constrained DTW Preserving Shapelets
Learning time-series representations that incorporate expert constraints while preserving Dynamic Time Warping relationships.
Published research
Shapelet Cluster Explanation
Explaining time-series clusters through shapelets, connecting learned representations to interpretable subsequences.
Published research
Neural vocoders beyond their training range
Evaluating how neural vocoders generalize to unseen pitch values rather than only testing within their training conditions.
03 / Publications
Published work
IEEE/ACS 22nd International Conference on Computer Systems and Applications (AICCSA)
Toward EEG-Free Seizure Detection: Evidence of EEG-ECG Synchronization
Constrained DTW Preserving Shapelets - CDPS
Constrained DTW Preserving Shapelets
No publications match these filters. Try another keyword or year.
04 / Experience
Research, teaching & practice
September 2025–present
Assistant Professor & Researcher
University of Sciences and Arts in Lebanon (USAL) · Lebanon
Teaching data structures, advanced data structures, introductory Java programming, and networks. Research on constrained clustering quality and diffusion-based learning for seizure detection.
2023–2025
Research Assistant
ICube Laboratory, University of Strasbourg · France
Active learning for incremental constrained clustering of time-series and remote-sensing data, building on CDPS and SCE.
2020–2023
PhD Researcher
ICube Laboratory, University of Strasbourg · France
Interpretable time-series representations integrating expert knowledge for clustering, using shapelets, contrastive learning, and Siamese networks.
2020–2024
Lecturer
University of Strasbourg · France
Linux tools and scripting; reading and discussing computer science research in English; coaching an AI hackathon.
2020
Research Intern · Speech Synthesis
GIPSA-LAB, Grenoble · France
Benchmarking how autoregressive, source-filter, flow, and GAN vocoders extrapolate to unseen pitch values.
2019
Telecommunication Engineering Intern
Ogero Center, Beirut · Lebanon
Telecommunication architecture and protocols.
05 / Teaching & supervision
From curiosity to research
Supporting students from their first programs to independent research.
Courses at USAL
- Introduction to Programming with Java
- Data Structures
- Advanced Data Structures
- Networks
Community Aware Coded Caching
Lebanese University · Master’s project
Supervision of a master’s student working on community-aware coded caching.
06 / Notes & ideas
Thinking out loud
Research notes, open questions, and thoughts on machine learning. Longer essays can also be shared on Medium.
A space for research notes and open questions. There are no posts here yet.
07 / Get in touch
Let’s explore a research question together.
For research collaborations, academic opportunities, or industry R&D, tell me what you are working on and where our interests connect.
alamourihusein@gmail.com