Inference in Statistical Relational AI

A tutorial on Statistical Relational AI at the ICCS 2019

24th International Conference on Conceptual Structures, July 1st-4th 2019, University Marburg

Overview

In recent years, a need for efficient inference algorithms on compact representations of large relational databases became apparent, e.g., in natural language understanding, machine learning, or decision making. This need has lead to advances in probabilistic relational modelling for artificial intelligence (also called statistical relational AI). Probabilistic relational models combine the fields of reasoning under uncertainty and modelling incorporating relations and objects in the vain of first-order logic.

Contents

  1. Introduction
  2. Probabilistic (relational) modeling
    • Models 
    • Inference problems
    • Inference algorithms
  3. Exact inference
    • Variable elimination based algorithms
      • Variable elimination (VE), lifted VE
      • Junction tree algorithm (JT), lifted JT
      • Temporal: Interface algorithm (IA), lifted dynamic JT
    • Weighted model counting based algorithms
      • Weighted model counting, weighted first-order model counting
      • Knowledge compilation (KC), lifted KC
      • Temporal: IA based
  4. Summary

Download Presentations

To appear