Profile
Maaz AHMAD
PhD Student
Maaz Ahmad's PhD research has focused on advancing the area of data-driven surrogate modeling and optimsation to aid industrial digitalisation. He has worked on developing a meta-learning-based surrogate selection framework to automate the process of selecting the best surrogate model for any given data set. The studies have also included a more fundamental analysis of different modeling techniques to compare and group similar performing surrogates into distinct families. He has also developed a surrogate-based optimisation framework based on a clustering algorithm, that does not require any derivative information. Hence, this derivative-free optimisation algorithm could potentially be used for optimising black-box systems with unknown form of the objectives. While his focus has been on relatively simpler problems involving bound-constraints and continuous decision variables, Maaz aims to extend the algorithm for more complex optimisation problems involving linear/non-linear constraints, and integer variables.
As part of his future works, Maaz wishes to explore improved modeling methods, possibly incorporating information of the system physics.
Past Members
Past Members
Research Interest
Key Publications
Ahmad, M., Karimi, I.A., 2021. Revised learning based evolutionary assistive paradigm for surrogate selection (LEAPS2v2). Computers & Chemical Engineering 152, 107385
Ahmad, M., Karimi, I.A., 2022. Families of similar surrogate forms based on predictive accuracy and model complexity. Computers & Chemical Engineering 163, 107845.
Conferences Attended & Corresponding Papers:
2020 Virtual AIChE Annual Meeting, November 16 – November 20, 2020. Oral Presentation: “Upgraded LEAPS2 for Surrogate Recommendation”
European Symposium on Computer Aided Process Engineering 2021, Istanbul, Turkey, June 6 – June 9, 2021
Conference Paper: Ahmad, M., Karimi, I.A., 2021. Upgraded Meta-Learning based Surrogate Selection Paradigm (LEAPS2v2), in: Türkay, M., Gani, R. (Eds.), Computer Aided Chemical Engineering. Elsevier, pp. 1099–1104.
2021 Virtual AIChE Annual Meeting, November 7 – November 19, 2021. Oral Presentation: “Families of Data-driven Surrogate Forms based on Accuracy and Complexity”
PSE 2021+, Kyoto, Japan, June 19 – June 23, 2022.
Conference Paper: M. Ahmad and I. A. Karimi, “Surrogate Classification based on Accuracy and Complexity,” in Computer Aided Chemical Engineering, Elsevier, 2022, pp. 1735–1740. doi: 10.1016/B978-0-323-85159-6.50289-X.
2022 AIChE Annual Meeting, November 13 – November 18, 2022. Oral Presentation: “Surrogate-based Optimization for Box-Constrained Black-Box Systems via k-means Clustering”