Ever think about a citywide network of electric vehicle (EV) fast-charging stations that suffers from rush-hour queues and volatile electricity prices?

Melike Yildiz Aktas has thought about this – a lot.

She has focused much of her research on building models that forecast arrivals and grid prices; assign chargers and set dynamic prices with priority service when needed; and route vehicles across stations using feedback control.

In practice, these models can reduce wait times, balance load across stations, and cut operating costs, all of which improve the electric vehicle driver experience while easing stress on the grid.

As a Ph.D. student in computer science at the Sanghani Center, she has extended her work on electric-vehicle (EV) charging networks to include data-rich social and public-service settings like education, healthcare language tools, and crime modeling, for example.

“I believe that when we have the power to make a difference, we have a responsibility to use it for good. That’s why I’ve chosen to contribute to meaningful change in the world,” said Yildiz Aktas, who is advised by Chang-Tien Lu.

Designing data-driven decision systems for urban infrastructure, her work combines machine learning (time-series forecasting, graph neural networks, temporal knowledge graphs, and large language models ) with operations and control to optimize resource allocation, pricing, and routing. 

She received both a bachelor’s and master’s degree in industrial engineering from TOBB University of Economics and Technology, Ankara, Turkey.

“Virginia Tech’s strength in engineering and its collaborative culture make it an ideal place for me to pursue my Ph.D.  As AI accelerates, my interest in computer science has grown. I hope to combine that training with my industrial engineering experience to tackle challenges in urban computing,” Yildiz Aktas said. “I’m especially drawn to the Sanghani Center because its active research portfolio and strong support align with my goals.”

She said that as part of the Sanghani Center community in Alexandria, she has found friends eager to team up on projects.  

In late August, she presented the paper, "MultiScale Spectral GNN for Fraud Detection," a collaboration with Mustafa Coskun at Ankara University, Turkey and Chang Tien- Lu,  at the International Conference on Advances in Social Network Analysis and Mining (ASONAM).

Papers from other conferences include:

·      "Time Series Forecasting with GCN-LSTM Based Unified Model for Product Demand Prediction" in proceedings of the 2024 IEEE International Conference on Big Data, Special Session: Machine Learning on Big Data (MLBD) 

·      "Enhancing School Success Prediction with FRC and Merged GNN" in proceedings of the 2024 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM)

·      "Synthetic Arabic Medical Dialogues Using Advanced Multi-Agent LLM Techniques" in proceedings of  The Second Arabic Natural Language Processing Conference (ArabicNLP), 2024

·      "UniMHe: Unified Multi Hyperedge Prediction A Case Study on Crime Dataset" in proceedings of the 2023 IEEE International Conference on Big Data, Special Session: Machine Learning on Big Data (MLBD)

 "Electric vehicle charging service operations: A review of machine learning applications for infrastructure planning, control, pricing and routing" was published in Renewable and Sustainable Energy Reviews in 2023.

“I aspire to become an academic just like the professors who inspired me and guided me through my education journey,” said Yildiz Aktas when asked about her long-term goal. 

She is projected to graduate in May 2028.