My name is Kimia Haghjooei and I hold an M.Sc. in Computer Science from Tarbiat Modares University, where I had the opportunity to work under the supervision of Dr. Mansour Rezghi. My thesis focused on adversarial examples and how they affect deep learning models. “Adversarial examples are subtle, imperceptible perturbations that expose the vulnerabilities of AI models and provide valuable insights. In fact, researchers can use adversarial examples to develop more robust models.
During my master’s, I worked on designing query-efficient black-box adversarial examples for video action recognition models. My research has led to two black-box adversarial attacks: the QEBB attack, which was published at ICPRAM 2024, and TenAdv, which has been submitted to the Elsevier journal Knowledge-Based Systems and is now under review.
Prior to my master’s, I earned my B.Sc. in Computer Science from Alzahra University, where I worked under Dr. Bita Shams on a recommender system using Jaccard and Pop Rank similarity metrics.
Currently, I work as an AI Engineer at Golrang Industrial Group, where I develop AI-driven solutions for real-time image and video processing and generative AI. Previously, I worked as a Computer Vision Engineer at Robokids, where I designed AI-powered interactive learning tools using computer vision for children’s education.
Research Interests:
- Adversarial Machine Learning – Exploring vulnerabilities in AI models and enhancing the robustness of these models
- Computer Vision & Video Analysis – Object detection, action recognition, sparse coding, pose estimation
Publications & Conferences:
- Haghjooei, K. & Rezghi, M. (2024). QEBB: A Query-Efficient Black-Box Adversarial Attack on Video Recognition Models Based on Unsupervised Key Frame Selection. Proceedings of the 13th International Conference on Pattern Recognition Applications and Methods (ICPRAM).
- Haghjooei, K. & Rezghi, M. (Submitted). TenAdv: A Tensor-Based Low-rank Black-Box Adversarial Attack on Video Classification Models.
- Conference Oral Presentation (ICPRAM 2024, Rome, Italy) – “QEBB: A Query-Efficient Black-Box Adversarial Attack on Video Recognition Models”
- Invited Speaker (GRCRAI 2024, Dubai, UAE) – “Enhanced Black-Box Adversarial Attack on Video Recognition Models Using NMF-Based Video Summarization”