As part of Al al-Bayt University's efforts to enhance its presence in international scientific forums and support scientific research and innovation in modern technology fields, the Department of Data Science and Artificial Intelligence at the Faculty of Information Technology participated in the First International Conference on Sustainable Innovation in Engineering and Technology (ICSIE 2026), organized by Al-Hijjawi Faculty of Engineering Technology at Yarmouk University, with the participation of researchers and academics from various countries worldwide.

The university was represented in this participation by student Mahmoud Obaidat from the Department of Data Science and Artificial Intelligence, in cooperation with Dr. Suheila Abu Oweida from the same department, by presenting a scientific paper in English titled:

"A Multi-Dataset Study of Task Sequencing and Its Data-Size-Dependent Crossover"

The research discussed the effect of training task order on improving the interpretability of artificial intelligence models, using knowledge distillation and explainable artificial intelligence techniques, with a focus on maintaining the models' predictive ability and enhancing the accuracy of explanations that clarify the impact of different features on model decisions.

The study relied on comparing three model training strategies, using the XGBoost model as a reference model, and SHAP techniques to determine the contribution of features to predictions, in addition to training a small neural network capable of simulating the outputs and explanations of the reference model. The experiments also included five public datasets in financial, medical, marketing, and statistical fields, aiming to study the relationship between the size of training data and task order and their impact on the quality of the resulting explanations.

The results showed that the order of training tasks affects the quality of explanations produced by the models, and that the most appropriate strategy varies depending on the size of the available data. Training that begins with learning explanations achieved better results when using small and medium data sizes in the expansion experiment, while joint task training achieved better results when using 20,000 samples. These results highlight the importance of designing training strategies that are appropriate for the nature and size of the data, thereby contributing to the development of more efficient and interpretable artificial intelligence models.

The study gains scientific and practical importance from its contribution to addressing one of the important challenges in artificial intelligence, which is achieving a balance between predictive efficiency and clarity of explanations, especially in applications that require understanding the reasons for automated decisions and the ability to review and evaluate their reliability. Its results also provide practical indicators that can be used when designing machine learning models that combine execution speed with maintaining explanation quality.

This participation reflects the level of interest that the Department of Data Science and Artificial Intelligence at Al al-Bayt University gives to encouraging applied scientific research and enabling students to participate in knowledge production and the development of technically based solutions, in addition to promoting research cooperation between academics and researchers from Jordanian universities.

Al al-Bayt University affirms its continued interest in developing the academic and research environment and supporting scientific initiatives that contribute to developing the capabilities of students and faculty members, and enhancing the university's presence in international scientific conferences, in line with its directions towards academic excellence, innovation, and digital transformation, and contributing to serving the community and keeping pace with rapid developments in the fields of data science and artificial intelligence.