Ontology-guided Semantic Composition for Zero-Shot Learning
This work addresses zero-shot learning for AI systems needing to generalize to unseen classes, but it appears incremental as it builds on existing ontology and embedding methods.
The authors tackled zero-shot learning by modeling class semantics with an OWL ontology and embedding it into a new framework, achieving effectiveness in primary experiments on animal image classification and visual question answering.
Zero-shot learning (ZSL) is a popular research problem that aims at predicting for those classes that have never appeared in the training stage by utilizing the inter-class relationship with some side information. In this study, we propose to model the compositional and expressive semantics of class labels by an OWL (Web Ontology Language) ontology, and further develop a new ZSL framework with ontology embedding. The effectiveness has been verified by some primary experiments on animal image classification and visual question answering.