This course aims to teach learners about Algorithmic Transparency via Quantitative Input Influence. The learning outcomes include understanding the importance of transparency in decision-making systems that employ machine learning, learning about Quantitative Input Influence (QII) measures to measure the influence of inputs on system outputs, and exploring the transparency-privacy tradeoff. The course covers topics such as learning systems, input correlation, marginal influence, and game theory. The teaching method involves theoretical explanations, experiments, and empirical validations with machine learning algorithms. This course is intended for individuals interested in understanding and improving the transparency of algorithmic decision-making systems.
Overview
Syllabus
Introduction
Learning Systems
Algorithmic Transparency
Input Correlation
Quantitative Input Influence
Marginal Influence
Game Theory
Examples
Related Work
Taught by
IEEE Symposium on Security and Privacy