Otmane Amel and Sédrick Stassin, researchers specializing in multimodal learning and explainability, recently presented their innovative research at the European Symposium on Artificial Neural Networks (ESANN) held from October 4-6, 2023. The event featured cutting-edge research in AI, where the two researchers showcased their work on multimodal learning algorithms aimed at enhancing explainability and performance across various applications. 🎉

Otmane and Sédrick presented two notable articles at ESANN 2023, each addressing different aspects of their multimodal learning approach:

• Multimodal Approach for Harmonized System Code Prediction tackles the challenges faced by customs representatives due to increasing e-commerce flows. Using a deep learning model that combines image and text data from customs declarations, the researchers introduced a novel MultConcat fusion method, achieving a top-3 and top-5 accuracy of 93.5% and 98.2%, respectively. 📊
  Read the full article here

• Similarity versus Supervision: Best Approaches for HS Code Prediction compares sentence embedding models based on semantic similarity with supervised models for predicting HS10 codes. Their results showcased the outstanding performance of the semantic similarity approach with a top-3 accuracy of 89% and a top-5 accuracy of 94.8%. 🔍
  Read the full article here

Amel and Stassin’s presentations at ESANN 2023 underscore the significant advancements in multimodal learning and the crucial role of explainable AI in practical applications. ESANN 2023 provided an excellent platform for these researchers to connect with other experts, fostering collaboration and innovation in the ever-evolving field of AI. 🤝🌐 AND they were taken on a visit of the beautiful belgian city of bruges at the end !