Learning With Relational Knowledge in the Context of Cognition, Quantum Computing, and Causality
MCML Authors
Abstract
Abstract
This dissertation explores the use of knowledge graphs, including semantic and episodic graphs, for representing static and evolving human knowledge, and proposes methods for improving knowledge inference. It introduces two quantum machine learning algorithms aimed at speeding up knowledge graph inference, demonstrating significant speedups over classical methods. Additionally, the work addresses causal inference in relational data, specifically in social networks, and proposes causal estimators using graph neural networks to estimate superimposed effects and optimize treatment assignments for network welfare. (Shortened.)
BibTeXKey: Ma20