The TensorDec Laboratory at the University of Trento is pleased to announce the masterclass

Nonnegative Low-Rank Approximations
Jeremy Cohen (CNRS)
5–9 October 2026, University of Trento

The masterclass will introduce nonnegative matrix and tensor approximations, combining theoretical foundations, optimization algorithms, and hands-on implementation. Applications will focus on music information retrieval, including automatic piano transcription and music segmentation, with methods relevant also to signal processing, image analysis, and machine learning.

The programme consists of five two-hour lectures, covering:

• Nonnegative matrix factorization and nonnegative tensor decompositions;

• Nonnegative least squares and numerical optimization;

• Practical implementation of factorization algorithms;

• Applications to music transcription and segmentation.

Lectures will take place at Povo 1 (Polo Ferrari), Trento, with the possibility of remote participation via Zoom for external attendants.

Registration is free but mandatory at the following FORM.

Full programme, schedule, and practical information:
Masterclass webpage

Please feel free to circulate this announcement among interested colleagues and students.

For information write to: 

alessandra.bernardi@unitn.it
elisa.postinghel@unitn.it
mima.stanojkovski@unitn.it