Argumentation and Machine Learning

Tutorial Description

This tutorial will give an introduction to the emerging field of research that combines argumentation and machine learning (ML). The goal is to exhibit works and ideas of how argumentation and ML come together in AI research and what the main purposes of combining the two are. In broad strokes, the tutorial will present how argumentation is used to improve and/or explain ML models and how ML is used to generate and compute argumentation frameworks, indicating where the combination of the two kinds of techniques seems to be more or less fruitful. In this regard, the tutorial will motivate and explain argumentative explainable AI as a topic of emerging importance. The tutorial will also offer a critical view on the lines of research, arguing, particularly, that scaling of any method combining argumentation and ML is crucial for successful applications and uptake. In summary, this tutorial will equip attendees with knowledge for contributing towards the integration of symbolic and data-driven approaches to AI, a highly active yet open-ended area of research, with great potential along numerous avenues.

This tutorial is aimed at any AI researcher interested in methods that combine the distinct lines of AI research of argumentation and ML. They may be interested in how argumentation can contribute to ML, or how ML can contribute to argumentation, or both. The tutorial will be self-contained, with requisite background on both argumentation and ML provided, though we expect attendees to be comfortable with quickly grasping the basics of either argumentation or ML.