Overview

The increasing popularity of machine learning (ML) models in various high-stakes domains such as finance, criminal justice and healthcare has resulted in a call for greater transparency into the mechanisms behind their predictions. Since transparency in algorithmic decision-making systems has undoubtedly become an ethical and sometimes legal obligation, there has been a recent surge of interest towards the field of algorithmic recourse. Algorithmic recourse aims to design mechanisms that provide actionable feedback about how to change the outcomes of ML models. For example, in the case of a ML model rejecting a loan application, an explanation that provides recourse could be: “To have your loan approved, you would need to increase your income by $10,000 per year”.

Although methods to provide algorithmic recourse are increasingly common in the ML literature, they are difficult to operationalize in practice. In this workshop, we aim to identify how to bridge the gap between theory and practice for algorithmic recourse. Specifically, we plan to facilitate workshop interactions that will shed light onto the following three questions:

• What are the practical, legal and ethical considerations that decision-makers need to account for when providing recourse?

• How do humans understand and act based on recourse explanations from a psychological and behavioral perspective?

• What are the main technical advances in explainability and causality in ML required for achieving recourse?

Through invited keynotes, an interactive panel discussion and contributed talks & posters, we will foster conversations that will help bridge the gaps arising from the interdisciplinary nature of algorithmic recourse and contribute towards the wider practical adoption of such methods.

Organizers

Manuel Gomez-Rodriguez, Amir-Hossein Karimi, Himabindu Lakkaraju, Ana Lucic, Stratis Tsirtsis, and Isabel Valera

For questions, contact us at recourse21@gmail.com.