Back to the program.
Peer review - Human activity recognition in smart meter data: What failed so far & next steps
Guillaume Levasseur
IRIDIA, Université Libre de Bruxelles
On 2023-04-26 at 15:00:00 (Brussels Time)

Abstract

Have you ever thrown millions of parameters to a problem without success? In this interactive seminar, I will give a gentle introduction to wonderful, real-world water data classification: feature spaces violating assumptions, six different machine learning approaches, and random output with 50% accuracy or less. After sharing the experimental setback to your greatest entertainment, I will present available options going forward. Through this seminar, I would like to collect your input and review of my work before researching further. Presentation: 30 minutes. Technical discussion: 30 minutes.