AIDA Short Course: “Learning to Quantify: Inferring Unbiased Estimators of Class Prevalence via Machine Learning”, 9,14,16 & 21 March 2023, 10:00-12:00 CET

Lecturer and Affiliation:

Alejandro Moreo, alejandro.moreo@isti.cnr.it

Fabrizio Sebastiani, fabrizio.sebastiani@isti.cnr.it

 

Host Institution: Italian National Council of Research (CNR), Institute for the Science and Technologies of Information (ISTI)

 

Content and organization: The goal of the course is to present an introduction to and a survey of the main approaches to Learning to Quantify (LtQ, also known as “quantification”, or “supervised prevalence estimation”). This task consists of training, from a set of data items labelled with class labels, a predictor that estimates, in a set of unlabelled data items, the prevalence values (i.e., relative frequencies) of the classes. Quantification thus stands to classification as aggregate data stand to individual data. While quantification may be trivially accomplished by training a standard classifier, classifying all the data items, counting the data items which have been assigned to each class, and then normalizing the resulting counts, it has been shown that this “classify and count” approach may be severely inaccurate, especially in scenarios characterized by “dataset shift”, i.e., by a substantial difference between the distribution of the items in the training data and the analogous distribution in the unlabelled data. This course will introduce the task of Learning to Quantify and its applications, discuss dataset shift, present supervised methods for training quantifiers, discuss evaluation measures and evaluation protocols for quantification, and assess the relative methods of different quantification methods. A hands-on session will also be offered, using a suite (called QuaPy) of tools for quantification implemented by the proposers using Python and Scikit-Learn.

 

Course Level: Postgraduate

 

Course duration: 8 hours (4 lectures, 2 hours each)

 

Course Type: Short course

 

Participation terms: Free for everybody.

Both AIDA and non-AIDA students are encouraged to participate in this short course.

 

If you are an AIDA Student* already, please:

 

Step (a): Register in the course by sending an email to the Course Lecturer alejandro.moreo@isti.cnr.it for your registration.

 

AND

 

Step (b): Enroll in the same course in the AIDA course link (TBA) using the ‘Enroll on this course’ button, so that this course enters your AIDA Certificate of Course Attendance.

 

If you are not an AIDA Student do only step (a).

 

*AIDA Students should have been registered in the AIDA system already (they are PhD students or PostDocs that belong only to the AIDA Members listed in this page: Members)

 

 

Lectures plan:

Thursday, March 9, 2023 – 10:00 to 12:00 CET

Tuesday, March 14, 2023 – 10:00 to 12:00 CET

Thursday, March 16, 2023 – 10:00 to 12:00 CET

Tuesday, March 21, 2023 – 10:00 to 12:00 CET

 

Language: English

 

Modality: online

 

Course link: https://learning2quantifyaida.github.io/course/

 

Both comments and pings are currently closed.

Comments are closed.

Design by 2b Consult