Fatemeh is a Ph.D. researcher in Computer Science and Information Systems at the University of Limerick. Her research interests lie in AI safety, deep generative models, and mathematical programming, ranging​ from​ theory​ to​ design​ to​ implementation. Her work investigates the robustness and out-of-distribution generalization of deep neural networks, while also exploring the capabilities and vulnerabilities of deep generative models, particularly GANs and diffusion models. In addition, she works on optimization problems related to dynamic ridesharing and transportation systems.

Publications:

Defending Against Frequency-Based Attacks with Diffusion Models, Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR) Workshops, 2025, pp. 3523-3533 https://openaccess.thecvf.com/content/CVPR2025W/AdvML/html/Amerehi_Defending_Against_Frequency-Based_Attacks_with_Diffusion_Models_CVPRW_2025_paper.html

Label Augmentation for Neural Networks Robustness, Proceedings of The 3rd Conference on Lifelong Learning Agents, PMLR 274:620-640, 2025 https://proceedings.mlr.press/v274/amerehi25a.html

Narrowing Class-Wise Robustness Gaps in Adversarial Training, ICLR 2025 Workshop on Foundation Models in the Wild https://openreview.net/forum?id=23zngyObBI

VF-Net: Robustness Via Understanding Distortions and Transformations, 2024 IEEE International Conference on Image Processing (ICIP) https://ieeexplore.ieee.org/abstract/document/10647346

Ridesharing: Harnessing Role Flexibility and HOV Integration, Proceedings of the 10th TRA Conference 2024 https://arxiv.org/abs/2504.05077

Ridesharing: The Impact of Traffic Restrictions and Role Flexibility, 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC), https://ieeexplore.ieee.org/abstract/document/10422162

My research expertise is in Computer vision and Deep Learning. I did my masters thesis on “Robust detection and Classification of vehicles in urban complex environment”. For this project our team built our own dataset and we implemented different detection and classification techniques on it. I got my first publication on my thesis topic. I presented my research paper in the Future technologies Conference which was held in San Francisco, USA. Then I started working on other two projects which are rice classification and Fingerprint recognition with teammates. We did 2 more publications on each topic and both were presented in an international conference. I am very interested in working on biomedical images in which these techniques can be used. Fortunately, I got a chance to select my research topic like that. I will be doing research on multiparametric identification and segmentation of breast cancer using DNN and Grammatical Evolution.

Supervisor: Professor Conor Ryan

Mehdi Yekrangi achieved his B.Sc. in Information technology from IUST (Iran University of Science and Technology) and his M.Sc. in E-commerce from QIAU (Qazvin Islamic Azad University), Mehdi joined the SFI Centre for Research Training in Artificial Intelligence. He has a great interest in the Financial Markets which interested him to start learning and working in CFD markets, dealing with various data produced daily in these markets from the very beginning of his undergraduate studies. During his undergraduate studies, he was mostly engaged in developing applications and websites; and due to his great interest in the finance domain, he developed an application with advanced capabilities for accounting affairs for his Bachelor’s graduation work. The abundant data in Social Networks and Financial Markets have always fascinated him to employ it gaining novel insights about sentiments and market behaviors. Therefore, he continued his Master’s studies focused on Natural Language Processing and Sentiment analysis in the Financial Markets. During his researches, he has worked on the relationship between social networks and currency market trends to develop a domain-specific lexicon for Sentiment Analysis in the Financial Markets which has resulted in a journal paper. Since his experiences in the Financial Markets are in close connection with Data Science, he has worked with various kinds of structured/unstructured data. He has applied various Data Mining and Machine Learning methods and techniques solving challenges and problems while improving Technical, Fundamental, and Sentimental approaches to enhance investment results. Mehdi is now working on the ontologies and developing trustworthy resources for Sentiment Analysis in the financial domain and employing them in deep learning architectures.

Supervisor: Dr. Nikola Nikolov & Professor Tiziana Margaria