Dr Zack Xuereb Conti

Research interests

Zack is passionate about multidisciplinary research that challenges traditional data-driven modelling approaches towards a more generalisable, interpretable and data-efficient machine learning that truly caters for dynamical engineering systems in real-world engineering applications. His current work seeks to improve the generalisability and interpretability of building energy forecasting models for assisting the decarbonisation of buildings, by combining well-known governing heat transfer equations together with unsupervised Reduced Order Modelling techniques in a subspace based- Domain Adaptation framework. Incorporating Physics into data-driven techniques leverages the governing structure of mechanistic knowledge while preserves mechanistic interpretability of the underlying dynamics. Zack is a co-founder of a popular online seminar series focusing on the advancement of Physics-enhanced Machine Learning with applications across Engineering fields, hosted at The Alan Turing Institute.

Previously, Zack's doctoral research focused on developing inference mechanisms to promote human intelligence at the interface of architecture and engineering in building design. More specifically, the work presented a novel bi-directional metamodeling approach using Bayesian networks to facilitate inverse prediction of the design space in parametric finite element analysis for structural engineering, and the interpretability of engineering responses from numerical analysis.