A Dual Machine-Learning Approach for Ensuring Authenticity and Sustainability of Agarwood Through Advanced Species Classification

Abstract

Precise identification of Aquilaria species is essential for maintaining the authenticity and sustainability of agarwood, a high-value resinous wood used in perfumery, traditional medicine, and religious rituals. Conventional identification methods often fail due to overlapping morphological and chemical characteristics. This study introduces a dual machine learning approach combining Self-Organizing Maps (SOM) for unsupervised feature selection with Artificial Neural Networks (ANN) for supervised classification. Essential oil profiles from four Aquilaria species (A. beccariana, A. malaccensis, A. crassna, and A. subintegra) were analysed, revealing δ-guaiene, 10-epi-γ-eudesmol, and γ-eudesmol as significant discriminative markers. The ANN model achieved 100% classification accuracy, demonstrating the effectiveness of this hybrid approach. This approach provides a robust and scalable solution for species authentication, supporting quality assurance and sustainable resource management within the agarwood industry.