Trusting SVM for piecewise linear CNNs
We present a novel layerwise optimization algorithm for the learning objective of Piecewise-Linear Convolutional Neural Networks (PL-CNNs), a large class of convolutional neural networks. Specifically, PL-CNNs employ piecewise linear non-linearities such as the commonly used ReLU and max-pool, and a...
Autors principals: | Berrada, L, Zisserman, A, Kumar, MP |
---|---|
Format: | Conference item |
Idioma: | English |
Publicat: |
Curran Associates
2017
|
Ítems similars
-
Trusting SVM for piecewise linear CNNs
per: Berrada, L, et al.
Publicat: (2016) -
An efficient hybrid model of CNNs and different kernels of SVM for brain tumor classification
per: Mohammed Bourennane, et al.
Publicat: (2023-10-01) -
Piecewise linear topology/
per: 456810 Hudson, J. F. P.
Publicat: (1969) -
Piecewise-linear and birational toggling
per: David Einstein, et al.
Publicat: (2014-01-01) -
Piecewise linear Wilson lines
per: Van der Veken Frederik F.
Publicat: (2015-01-01)