AUTHOR=Jiang Weizheng , Chen Zhibo , Zhang Haiyan , Li Juhu TITLE=MelSPPNET—A self-explainable recognition model for emerald ash borer vibrational signals JOURNAL=Frontiers in Forests and Global Change VOLUME=Volume 7 - 2024 YEAR=2024 URL=https://www.frontiersin.org/journals/forests-and-global-change/articles/10.3389/ffgc.2024.1239424 DOI=10.3389/ffgc.2024.1239424 ISSN=2624-893X ABSTRACT=To achieve early and reliable monitoring of wood borers, which are highly concealed, with long lag times and significant damage to forests, we selected the representative emerald ash borer larvae feeding sound as the research object. We designed a self-explainable model called MelSPPNET to extract prototypes from input vibration signals and obtain the most representative audio segments as the basis for model recognition. We collected the feeding vibration signals of emerald ash borer larvae and some typical outdoor noise using a detector. In the experiment, MelSPPNET achieved comparable accuracy to its similar uninterpretable counterpart network and provided interpretability that these networks do not possess. We designed an explanatory evaluation index for the case-based self-explainable model and demonstrated that MelSPPNET has good interpretability. MelSPPNET can provide accurate and reliable technical support for the recognition of emerald ash borer larvae. The work of this study is limited to one type of pest, and future experiments will focus on the applicability of this network to the recognition of other vibration signals.