Date: | 2024, March 4 |
Time: | 10:30 a. m. |
Place: | G29-018 |
Author: | Memariani, Adel |
Title: | Box Embeddings for Learning Chemical Hierarchies |
Knowledge graphs are great at organizing data and inferring additional facts through logical reasoning, but often dealing with uncertainty is beyond their capabilities. Alternatively, knowledge graph embeddings provide a promising method for predicting plausible information, handling uncertainty, and addressing missing relations. Knowledge graph embeddings are typically acquired through the training of deep neural networks. However, these embeddings are not explainable. In this presentation, we present a geometric embedding technique known as box embedding and demonstrate how visualizing the embedding space can offer interpretability.