Accepted paper at NeurIPS 2020 Workshop QTNML 2020
A Variational Quantum Circuit Model for Knowledge Graph Embeddings
06.11.2020
Authors
Yunpu Ma, Volker Tresp
1st Workshop at the 34th Conference on Neural Information Processing Systems (NeurIPS 2020): Quantum tensor networks in machine learning (QTNML 2020),
11th December 2020, Virtual
Abstract
Can quantum computing resources facilitate representation learning? In this work, we propose the first quantum Ansatz for statistical relational learning on knowl- edge graphs using parametric quantum circuits. We propose a variational quantum circuit for modeling knowledge graphs by introducing quantum representations of entities. In particular, latent representations of entities are encoded as coefficients of quantum states, while predicates are characterized by parametric gates acting on the quantum states. We show that quantum representations can be trained ef- ficiently meanwhile preserving the quantum advantages. Simulations on classical machines with different datasets show that our proposed quantum circuit Ansatz and quantum representations can achieve comparable results to the state-of-the-art classical models, e.g., RESCAL, DISTMULT. Furthermore, after optimizing the models, the complexity of inductive inference on the knowledge graphs can be reduced with respect to the number of entities. pdf.