Semi supervised identification of numerically simulated pile defects using graph label prppagation Semi supervised identification of numerically simulated pile defects using graph label prppagation Αφίσα σε Συνέδριο Conference Poster 2015-10-132015enA graph-based detector is utilized for the identification of structural defects in piles. The approach utilize semi supervised learning assumptions, which allow the exploitation of the unlabeled data to produce accurate defect identifications. A coupled finite element and scaled boundary finite element method approach is used to simulate the pile and its surrounding soil. The proposed defect detection system provides information about the type and the placement of the defect(s), given the surface's oscillation patterns.http://creativecommons.org/licenses/by/4.0/8th GRACM International Congress on Computational Mechanics Eftychios Protopapadakis Stavroulakis Georgios Σταυρουλακης Γεωργιος Marco Schauer Anastasios Doulamis Jens Böhrnsen Sabine Langer