|
The shallow learning quest| old_uid | 16801 |
|---|
| title | The shallow learning quest |
|---|
| start_date | 2018/11/20 |
|---|
| schedule | 14h-15h30 |
|---|
| online | no |
|---|
| location_info | salle de séminaire |
|---|
| summary | Successful deep neural networks are systematically trained via the end-to-end back-propagation algorithm. In a recent work, we challenge this optimization procedure. We show it is possible to train layers sequentially, by using ad-hoc auxiliary classifiers. For instance, one can obtain AlexNet performances on Imagenet by training successively a sequence of 1-hidden layer CNNs. Using deeper auxiliary classifier, we exhibit VGG performances. We propose an empirical analysis through the scope of progressive linear separation. Applications are discussed. |
|---|
| responsibles | Grandjean |
|---|
| |
|