Výsledky vyhledávání

Report

Comparing Dynamics: Deep Neural Networks versus Glassy Systems

Autor: Baity-Jesi, M., Sagun, L., Geiger, M., Spigler, S., Arous, G. Ben, Cammarota, C., LeCun, Y., Wyart, M., Biroli, G.

Publikováno v: PMLR 80:324-333, 2018; Republication with DOI (cite this one): J. Stat. Mech. (2019) 124013

We analyze numerically the training dynamics of deep neural networks (DNN) by using methods developed in statistical physics of glassy systems. The two main issues we address are (1) the complexity of the loss landscape and of the dynamics within it,

Externí odkaz: http://arxiv.org/abs/1803.06969

Zobrazit plný text záznamu

Akademický článek

Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.

Adversarially Regularized Autoencoders

Autor: Zhao, J., Kim, Y., Zhang, K., Alexander Rush, Lecun, Y.

Publikováno v: Scopus-Elsevier

Deep latent variable models, trained using variational autoencoders or generative adversarial networks, are now a key technique for representation learning of continuous structures. However, applying similar methods to discrete structures, such as te

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::6f5a1bf6520087ac50666b8902055bfe

Zobrazit plný text záznamu

Tunable Efficient Unitary Neural Networks (EUNN) and their application to RNNs

Autor: Jing, L., Shen, Y., Tena Dubcek, Peurifoy, J., Skirlo, S., Lecun, Y., Tegmark, M., Solja I, M.

Publikováno v: Scopus-Elsevier

Using unitary (instead of general) matrices in artificial neural networks (ANNs) is a promising way to solve the gradient explosion/vanishing problem, as well as to enable ANNs to learn long-term correlations in the data. This approach appears partic

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::1652b7f1ff8c682c19d90bf01cae7b59
http://arxiv.org/abs/1612.05231

Zobrazit plný text záznamu

Akademický článek

Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.

Indoor Semantic Segmentation using depth information

Autor: Couprie, C., Farabet, C., Laurent Najman, Lecun, Y.

Publikováno v: Proceedings of the International Conference on Learning Representations
First International Conference on Learning Representations (ICLR 2013)
First International Conference on Learning Representations (ICLR 2013), May 2013, Scottsdale, AZ, United States. pp.1-8
Scopus-Elsevier

This work addresses multi-class segmentation of indoor scenes with RGB-D inputs. While this area of research has gained much attention recently, most works still rely on hand-crafted features. In contrast, we apply a multiscale convolutional network

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::a6e2f3736c2ed3d7335174a1c269c901
https://hal.archives-ouvertes.fr/hal-00805105

Zobrazit plný text záznamu

Semantic Road Segmentation Via Multi-Scale Ensembles of Learned Features

Autor: Alvarez, J.M., LeCun, Y., Gevers, T., Lopez, A.M., Fusiello, A., Murino, V., Cucchiara, R.

Publikováno v: ResearcherID
Computer Vision – ECCV 2012. Workshops and Demonstrations ISBN: 9783642338670
ECCV Workshops (2)
Computer Vision – ECCV 2012 : Workshops and Demonstrations: Florence, Italy, October 7-13, 2012: proceedings, 2, 586-595

Semantic segmentation refers to the process of assigning an object label (e.g., building, road, sidewalk, car, pedestrian) to every pixel in an image. Common approaches formulate the task as a random field labeling problem modeling the interactions b

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::c5fb9ce06794da7a2754aab66cf8ff7a
https://hdl.handle.net/11245/1.379990

Zobrazit plný text záznamu

Vyhledávací nástroje:

Upřesnit hledání