We consider a structured multi-label prediction problem where the labels are organized under implication and mutual exclusion constraints. A major concern is to produce predictions that are logically consistent with these constraints. To do so, we formulate this problem as an embedding inference problem where the constraints are imposed onto the embeddings of labels by geometric construction. Particularly, we consider a hyperbolic Poincaré ball model in which we encode labels as Poincaré hyperplanes that work as linear decision boundaries. The hyperplanes are interpreted as convex regions such that the logical relationships (implication and exclusion) are geometrically encoded using the insideness and disjointedness of these regions, respectively. We show theoretical groundings of the method for preserving logical relationships in the embedding space. Extensive experiments on 12 datasets show 1) significant improvements in mean average precision; 2) lower number of constraint violations; 3) an order of magnitude fewer dimensions than baselines.
Bo Xiong (University of Stuttgart)
Michael Cochez (VU Amsterdam)
I am an Assistant Professor in the Learning and Reasoning group at the Vrije Universiteit Amsterdam.
Mojtaba Nayyeri (University of Stuttgart)
Steffen Staab (University of Stuttgart)
More from the Same Authors
2022 : Explanation Shift: Detecting distribution shifts on tabular data via the explanation space »
Carlos Mougan · Klaus Broelemann · Gjergji Kasneci · Thanassis Tiropanis · Steffen Staab
2023 Poster: Adapting Neural Link Predictors for Efficient Complex Query Answering »
Erik Arakelyan · Pasquale Minervini · Daniel Daza · Michael Cochez · Isabelle Augenstein
2022 Poster: Pseudo-Riemannian Graph Convolutional Networks »
Bo Xiong · Shichao Zhu · Nico Potyka · Shirui Pan · Chuan Zhou · Steffen Staab