Dongyu “Don” Han / Research engineer
I’m a research engineer with a computer science background. I study how machine learning can draw reliable conclusions from limited measurements—and how computation can help us decide what to measure next.
About me
I’m Dongyu Han, or Don. I studied Computer Science & Technology for my BEng, then completed an MPhil at HKUST, where I applied machine learning to sensing problems in water networks. I now work on algorithms for pipeline condition assessment.
Across software engineering and applied research, I keep returning to the same questions: what can we learn from scarce data, how can we understand a model’s decision, and which measurements are worth making?
Background
Hyele Limited · Hong Kong Science Park
Develop machine-learning algorithms for real-time pipeline condition assessment, from transient-signal processing and model validation to neural-network inference and sensor optimization.
HKUST · Civil and Environmental Engineering
Built reproducible simulation-to-field experiments for leak diagnosis and a contrastive, differentiable framework for sparse sensor selection.
Yuzhi Tech Co., Ltd. · Beijing
Built Java/Spring Boot services and integrated ML inference into a production digital-human system; contributed to a secondary-market knowledge graph project.
Research in practice
Water distribution networks gave me a concrete setting for these questions. A leak changes how transient pressure waves travel and reflect, yet labeled field measurements are scarce and sensors cannot go everywhere. My two research threads ask how to transfer knowledge from simulation to measured responses, and how to choose a useful subset of sensor locations.
01 Learn across the gap
We can simulate many leak scenarios, but collect only a small number of labeled transient pressure responses in the field. Simulations give a model a starting point; measured responses let us test and adapt it to the physical network.
I fine-tuned models trained on simulated transients using limited experimental measurements. I then used Integrated Gradients to inspect which sensor–time regions contributed most to their predictions.
Transfer learning · Interpretable leak diagnosisCan features learned from simulated transients still help when the model meets measured responses?
Transformer model · 17 experimental scenarios
Mean absolute error on this evaluation; lower is better. These are best-model results, not a general field-performance guarantee.
02 Choose where to measure
Even a strong model is limited by what its sensors record. In a larger network, adding sensors at every node is impractical; each location has to earn its place.
I built a method that learns which subset of locations is most useful for distinguishing leak events. The network below shows the branches and nodes; the changing selection probabilities show the model making that choice during training.
Contrastive learning · Sensor selection
One simulated Q‑Leak setting · 8 sensors · SNR = 0 dB
At some lower-SNR, tighter-budget settings, the MI/MaxDet baseline performs better. This comparison uses simulated responses, not field measurements.
Publication
The simulation-to-field study is published in Water Research. If you want the methods, assumptions, and full results, start here.
Water Research · 2026
Dongyu Han, Oussama Choura, Muhammad Waqar, Moez Louati, Dai Linhan & Mohamed Salah Ghidaoui
Read the paperIndependent project
A web app for exploring research papers. Upload a PDF, ask a question, inspect the source passages behind the answer, and browse the paper’s extracted relationships.
Explore EvidenceGraph