(2023). Their aim is to classify a sample of curves into homogeneous groups while jointly detecting the most informative portions of the domain in which the functions are observed. Extending the approach introduced by James and Sugar (2003), the log-likelihood is penalized using both a ...