Lien | https://s2d-olad.github.io/papers/submission-13.pdf |
---|
Auteur | Rechercher : Al-Digeil, Muhammad1; Rechercher : Grinberg, Yuri1; Rechercher : Kamandar Dezfouli, Mohsen2; Rechercher : Melati, Daniele; Rechercher : Schmid, Jens H.2; Rechercher : Cheben, Pavel2; Rechercher : Janz, Siegfried2; Rechercher : Xu, Dan-Xia2 |
---|
Affiliation | - Conseil national de recherches du Canada. Technologies numériques
- Conseil national de recherches du Canada. Électronique et photonique avancées
|
---|
Format | Texte, Article |
---|
Conférence | Ninth International Conference on Learning Representations 2021, May 3-7, 2021, Virtual Only |
---|
Résumé | Principal Component Analysis (PCA) provides reliable dimensionality reduction (DR) when data possesses linear properties even for small datasets. However, faced with data that exhibits non-linear behaviour, PCA cannot perform optimally as compared to non-linear DR methods such as AutoEncoders. By contrast, AutoEncoders typically require much larger datasets for training than PCA. This data requirement is a critical impediment in applications where samples are scarce and expensive to come by. One such area is nanophotonics component design where generating a single data point might involve running optimization methods that use computationally demanding solvers. We propose Guided AutoEncoders (G-AE) of nearly arbitrary architecture which are standard AutoEncoders initialized using a numerically stable procedure to replicate PCA behaviour before training. Our results show this approach yields a marked reduction in the data size requirements for training the network along with gains in capturing non-linearity during dimensionality reduction and thus performing better than PCA alone. |
---|
Date de publication | 2021-05 |
---|
Maison d’édition | International Conference on Learning Representations |
---|
Autre format | |
---|
Langue | anglais |
---|
Publications évaluées par des pairs | Non |
---|
Exporter la notice | Exporter en format RIS |
---|
Signaler une correction | Signaler une correction (s'ouvre dans un nouvel onglet) |
---|
Identificateur de l’enregistrement | 1e0df489-1401-4852-b02f-9486d0f5b46b |
---|
Enregistrement créé | 2022-05-30 |
---|
Enregistrement modifié | 2023-07-27 |
---|