Bridging the Gap Between AI and Reality • Rhodes, Greece
Time: Saturday, 1.11
Room: Room B
Authors: Amandeep Singh, Stephen Ryan, Tiziana Margaria, Kyriakos Kourousis
Abstract: The mechanical characterisation of additively manufactured metals, such as maraging steel 300, is often constrained by sparse experimental data, owing to the high cost and complexity of specimen preparation across varying build orientations and heat treatment configurations. In such contexts, data-driven modelling faces extreme sparsity and noise. These are conditions under which classical machine learning techniques are frequently ill-suited. This research introduces a domain-grounded approach to automating interpolation workflows for plastic anisotropy and mechanical properties, tailored to the nuanced constraints of additive manufacturing experiments. The resulting framework helps materials researchers to iteratively augment sparse datasets and refine their experiments. By highlighting automation and reproducibility in the absence of large-scale data, this research offers a viable alternative to black-box modelling by preserving the domain context while enabling responsible data-centric practices in materials science.