Dr Taha Mansouri
School of Science, Engineering & Environment
Current positions
Lecturer in AI
Biography
Taha is an AI, Computer Vision, and Deep Learning specialist with a dual PhD in Artificial Intelligence and Deep Learning from the University of Salford (UK) and Information Technology Management from Allameh Tabataba'i University (Iran).
He brings his 14+ years of industry experience to enrich his academic career, which began in 2012. Since joining the University of Salford as a Research Associate in 2019 and later as a Lecturer in AI in 2022, Taha's research focuses on applying AI solutions to real-world problems. He's a specialist in developing explainable deep learning for transparency, computer vision models, and exploring the ethical considerations of Large Language Models. Taha's impactful research, published in esteemed journals, demonstrates the depth and significance of his contributions to the field of AI.
Taha has developed techniques to make deep learning models more transparent including:
https://ieeexplore.ieee.org/abstract/document/9801817
https://onlinelibrary.wiley.com/doi/full/10.1111/exsy.13316
He has also made significant contributions to optimization algorithms, developing fundamental model-free approaches:
https://www.sciencedirect.com/science/article/abs/pii/S0950705118304854
https://www.sciencedirect.com/science/article/abs/pii/S0957417410010419
https://www.sciencedirect.com/science/article/abs/pii/S1568494610001122
Taha's implemented codes can be explored in the following link:
https://github.com/tahamsi
Taha's current research focuses on measuring bias and fairness in complex AI solutions, including large language models and computer vision systems. This vital work ensures the responsible development and deployment of AI.
Taha's editorial experience includes:
Guest Editor: "Blockchain: Applications, Challenges, and Solutions" https://www.mdpi.com/journal/futureinternet/special_issues/B_ACS
Current Guest Editor:
"Application of Sensor Technologies in Livestock Farming" for the prestigious journal Agriculture (Impact Factor 3.6). https://www.mdpi.com/journal/agriculture/special_issues/sensors_application
"Machine Vision Solutions and AI-Driven Systems in Agriculture."
https://www.mdpi.com/2076-3417/13/5/2879
He currently serves on the:
Conference Organizing Committee and Programme Committee of the 35th Annual Conference of the International Information Management Association.
Track Chair for Ethics in Digital, AI, Big Data, Data Science, and Marketing Science.
https://iima.org/wp/call-for-papers/
Areas of Research
In Explainable AI (XAI), Taha strives to make deep learning models more transparent. He explores techniques like instance-wise feature selection and explanation by simplification to improve understanding of model decisions. Additionally, Taha is investigating explanation by text generation, leveraging NLP for clearer explanations.
Within Computer Vision (CV), Taha delves into various tasks like image segmentation, classification, and object detection. Beyond that, He is fascinated by the potential of image-to-image translation and image generation using generative models. Taha is also interested in foundation models in CV.
Building specialized large language models (LLMs) for real-world use cases is another key focus. This includes exploring how LLMs can be effectively integrated with other AI models, such as computer vision systems and how to fine-tune LLMs through prompt tuning for vision-related tasks
Finally, He is passionate about ensuring AI Fairness. His research involves methods to measure and evaluate bias in AI models, with the goal of developing bias mitigation solutions. This could involve incorporating additional modalities like text or sensor data during model training.
Areas of Supervision
My research focus is on AI, particularly in Computer Vision, LLMs, and broader Multimodal Large AI Models. I'm passionate about exploring how AI can be applied to real-world problems and how it can be ethical.
If you share my interest in AI and possess strong programming and mathematical skills, I'd be happy to discuss potential research collaborations.
Taha has led the following modules:
Data Structures and Algorithms (DSA): Providing a strong foundation in computational thinking and problem-solving techniques.
Machine Learning and Data Mining (MLDM): Equipping students with the skills to analyze large datasets and extract meaningful insights.
He takes a leadership role, curriculum development, and delivering new modules, including:
Big Data Tools and Techniques (BDTT) (MSc): Equipping students with the skills to handle and analyze large and complex datasets.
Principles and Design of IoT Systems (PDIoT) (MSc): Providing students with the knowledge to design and develop Internet of Things (IoT) systems, in particular Computer Vision and Remote Sensing.
Taha is a Fellow of the Higher Education Academy (FHEA) and actively engages in professional development.