HACR - Hybrid Architecture for Concept Reasoning

A hybrid architecture for learning and reasoning about the concepts of ‘ holding’ and ‘entering’ from the TVQA+ dataset. Combines power of statistical learning (pre-trained neural models) with symbolic learning and reasoning ( Inductive Learning system ILASP and reasoning model ASP).

Thesis: HACR - Hybrid Architecture for Concept Reasoning

Tools: Python, Pytorch, ILASP, Scikit-learn

Keywords: Neuro-symbolic Learning Pipeline, Inductive Logic Programming