Learning Comprehensible Relational Features to Distinguish Subfossil Decapod Crustacean Dactyls

Our research explores the application of Inductive Logic Programming to a new domain involving decapod crustacean claws. We find that we can distinguish dactyl shapes by automatically extracting relational features that describe their underlying spatial structure. We first use medial axis techniques to find the shock graph of each dactyl outline, which is then converted into a first-order logic representation capturing the connections, distances and angles between the nodes in this graph. We then use Aleph to find relational classification rules based on the shock graph representations. These relational rules provide a concise and human-understandable way to describe the morphological differences between closely related decapods, and can be seen as a first step to creating automatically learned quantitative taxonomic keys.