Training a neural network for data reduction and better generalization: towards intelligent artificial intelligence.
Sylvain Sardy (Prof. associé, Université de Genève)
| When |
Apr 08, 2025
from 01:00 to 02:00 |
|---|---|
| Where | M7 101 |
| Attendees |
Sylvain Sardy |
| Add event to calendar |
vCal
iCal
|
Sylvain Sardy
Title: Training a neural network for data reduction and better generalization: towards intelligent artificial intelligence.
Abstract: The motivation for sparse learners is to compress the inputs (features) by selecting only the ones needed for good generalization. A human scientist can then give an intelligent interpretation to the few selected features. If genes are the inputs and cancer type is the output, then the selected genes give the cancerologist clues on what genes have an effect on certain cancers. LASSO-type regularization leads to good input selection for linear associations, but very few attempts have been made for artificial neural networks. A stringent but efficient way of testing whether a feature selection method works is to check if a phase transition occurs in the probability of retrieving the relevant features. Our method achieves just so thanks to a good optimization scheme (FISTA and warm start) and a good selection of a single regularization parameter.
Website: https://scholar.google.fr/citations?user=64yVlTUAAAAJ&view_op=list_works&sortby=pubdate
In Room M7 101, 1st floor, Monod campus, ENSL.