Zahed Rahmati
Amirkabir University of Technology, University of Waterloo, University of Victoria
Research areas: Computer Science · Computer Graphics and Computer-Aided Design · Computational Geometry and Mesh Generation · Complexity and Algorithms in Graphs · Data Management and Algorithms
0000-0002-6942-3453
About
Dr. Zahed Rahmati is an Assistant Professor at Amirkabir University of Technology (AUT), specializing in graph machine learning, knowledge graphs, algorithms, and computational geometry. His research bridges rigorous theoretical foundations with modern data-driven methodologies, contributing to both core algorithm design and applied artificial intelligence.
Dr. Rahmati began his academic career with influential work in computational geometry and kinetic data structures. He developed efficient algorithms for dynamic proximity problems, Euclidean minimum spanning trees of moving points, and nearest neighbor structures. His research on kinetic closest pair problems, reverse k-nearest neighbors, and geometric spanners—particularly the development of Emanation Graphs—has been widely cited. In collaboration with researchers such as Timothy M. Chan, he also contributed to improved approximation algorithms for the discrete Fréchet distance and clustering-based approaches to kinetic problems.
In recent years, his work has expanded into graph machine learning and deep learning on structured data. He has published on dynamic graph attention networks for stock prediction, centrality-aware deep learning for portfolio management, relational graph convolutional networks for sentiment analysis, and interpretable counterfactual explanations. His interdisciplinary research also includes drug–target interaction prediction, medical data analysis such as breast cancer classification, and spatio-temporal dust emission forecasting using graph-based models.
Dr. Rahmati’s publications appear in respected venues including Computational Geometry, Algorithmica, Information Processing Letters, Graphs and Combinatorics, and Computers in Biology and Medicine. His body of work reflects a consistent effort to integrate theoretical graph algorithms with practical machine learning systems, advancing both mathematical foundations and intelligent data-driven applications.
Employment
-
Amirkabir University of Technology Assistant Professor2016 - Present
-
University of Waterloo Postdoctoral Researcher2014 - 2016
Education
- 2011 - 2014
Projects & Funding
Projects & funding information is unavailable.
Publications (43)
- HBPP: A transformer-inspired architecture for physics-informed learning of parametric PDEs Save
- Towards Standardized Evaluation of Drug-Target Interaction Models: A Survey of Explainability Methods and Metrics Save
- Dust emission forecasting using time series spatial data and machine learning algorithms Save
- Graph centrality and deep learning for enhanced portfolio management Save
- Drug-target interaction/affinity prediction: Deep learning models and advances review Save
- On Enhancement of Text Classification and Analysis of Text Emotions Using Graph Machine Learning and Ensemble Learning Methods on Non-English Datasets Save
- Emanation Graph: A Plane Geometric Spanner with Steiner Points Save
- RAC drawings in subcubic area Save
- Simplified Emanation Graphs: A Sparse Plane Spanner with Steiner Points Save
- Simplified Emanation Graphs: A Sparse Plane Spanner with Steiner Points Save
- Kinetic k-Semi-Yao graph and its applications Save
- Simplified Emanation Graphs: A Sparse Plane Spanner with Steiner Points Save
- A Clustering-Based Approach to Kinetic Closest Pair Save
- An improved approximation algorithm for the discrete Fréchet distance Save
- Emanation Graph: A New t-Spanner Save
- Emanation graph: A new t-spanner Save
- Width and Bounding Box of Imprecise Points Save
- Width and bounding box of imprecise points Save
- Approximating the minimum closest pair distance and nearest neighbor distances of linearly moving points Save
- A Clustering-Based Approach to Kinetic Closest Pair Save