Best Approaches for Customs Fraud Detection

The rapid growth of e-commerce has placed considerable pressure on customs
representatives. Artificial intelligence (AI) systems have emerged as a promising approach
to minimize the risks faced in the customs domain. In this presentation, we introduce various
approaches that perform well for customs fraud detection: first, unsupervised approaches to predict a fraud label, and second, approaches to predict Harmonized System (HS) codes, a crucial element for an accurate customs declaration. Finally, we investigate the
performance of such models with the multimodal information at our disposal