Conformal learning – prediction with accuracy guarantees (theory & practice)

Course Description

This course is designed for university students of all levels and researchers interested in Machine Learning. The goal of this course is to present the paradigm of Conformal Learning that allows transforming any existing predictor into a predictor with accuracy guarantees. This course will guide you through the following topics: 1) Theory of conformal prediction and how it can guarantee the accuracy of, for example, 95%; 2) Conformal regression; 3) Conformal classification; 4) Examples of application.

Course tools

• Python
• Jupyter notebooks

Prerequisites

• python programming (intermediate level),
• machine-learning algorithms (beginner),
• statistics (intermediate level)

Level of complexity of course Intermediate

Lecturer

Dr. Marharyta Aleksandrova The University of Luxembourg, Postdoc After completing a Master’s degree from the National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Marharyta started her research path by pursuing a joint doctorate program between France (University of Lorraine) and Ukraine (Igor Sikorsky KPI). Currently, she holds the position of postdoctoral researcher at the University of Luxembourg. During these years, she has been exposed to various applications of Data Science and worked on several European and national research projects. Fields of interests: Data analysis, Theory of machine learning, Causal learning, Conformal learning, Recommender systems, Application of Data Mining to security Contacts: marharyta.aleksandrova@gmail.com