A deep neural network model for hazard classification

Joseph N. Wilson; Ferit Toska; Maksim Levental; Peter J. Dobbins

doi:10.1117/12.2535681

Translator Disclaimer

19 September 2019 A deep neural network model for hazard classification

Joseph N. Wilson, Ferit Toska, Maksim Levental, Peter J. Dobbins

Author Affiliations +

Proceedings Volume 11169, Artificial Intelligence and Machine Learning in Defense Applications; 1116903 (2019) https://doi.org/10.1117/12.2535681
Event: SPIE Security + Defence, 2019, Strasbourg, France

Abstract

Hazard learning algorithms employing ground penetrating radar (GPR) data for purposes of discrimination, detection, and classification suffer from a pernicious robustness problem; models trained on a particular physical region using a given sensor (antenna system) typically do not transfer effectively to diverse regions interrogated with differing sensors. We implement a novel training paradigm using region-based stratified cross-validation that improves learning induction across disparate data sets. We test this training paradigm on a novel deep neueral network architecture (DNN) and report empirical results from testing/training on data collected from multiple sites. Furthermore, we discuss the relationship between penalty loss and evaluation metrics.

Conference Presentation

Citation Download Citation

Joseph N. Wilson, Ferit Toska, Maksim Levental, and Peter J. Dobbins "A deep neural network model for hazard classification", Proc. SPIE 11169, Artificial Intelligence and Machine Learning in Defense Applications, 1116903 (19 September 2019); https://doi.org/10.1117/12.2535681

ACCESS THE FULL ARTICLE