Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 66: Topical Session (Symposium MM): Big Data in Materials Science - Managing and exploiting the raw material of the 21st century
MM 66.2: Vortrag
Donnerstag, 15. März 2018, 18:00–18:15, H 0107
Subgroup Discovery for Finding Local Patterns in Materials Data — •Mario Boley1, Bryan R. Goldsmith2, Christopher Sutton3, Jilles Vreeken1, Matthias Scheffler3, and Luca M. Ghiringhelli3 — 1Max-Planck-Institut für Informatik, Saarbrücken — 2University of Michigan, Ann Arbor — 3Fritz-Haber-Institut der Max-Planck-Gesellschaft, Berlin
We establish that subgroup discovery (SGD), a form of local pattern discovery for labeled data, can help find interpretable descriptors from materials-science data obtained by first-principles calculations. In contrast to global modelling algorithms, SGD finds descriptions of subpopulations in which, locally, the target property takes on an interesting distribution. First, we formulate the SGD algorithm for applications in scientific domains. Subsequently, SGD is applied to configurations of neutral gas-phase gold clusters to discern general and interesting patterns between their geometrical and physicochemical properties. For example, SGD uncovers that van der Waals interactions within gold clusters are linearly correlated with their radius of gyration and are weaker for planar clusters than for nonplanar clusters. Moreover, we explore SGD for finding descriptors that predict both the formation and bandgap energies of transparent conducting oxides as well as descriptors that classify the octet binary semiconductors as either rock salt or zincblende; in both settings using only information of their chemical composition. Lastly, an efficient optimal solver using branch-and-bound is developed for dispersion-corrected objective functions to facilitate the discovery of interpretable subgroups.