CNNs @ 3IT
Victor Schmidt & Simon Verret
Mila
3IT - Université de Sherbrooke
14 Novembre 2019
Hier
MLP
Loss
Backprop
Aujourd'hui
Convolutions
Réseaux convolutionnels
GANs
Motivation
Classification
Semantic Segmentation
Detection
Instance Segmentation
Image Generation
Image to Image Translation
How to deal with Images?
Filters
Stride = 2
Padding = 1
Convolutional layers
Some more convolutions
Padding = same
Dilated
Transposed
Small translations: pooling
In practice
Convolutional
Neural
Networks
Multi-channel convolutions
Normalizing the inputs
A first network: LeNet
LeCun, 1998
Stabilising training: Batch Norm
Allows the training of deeper networks
Less sensitive to initialization
Larger learning rates can be used
Faster and more stable convergence
Ioffe, 2015
Updated LeNet
Everything's differentiable!
Some more classification
ImageNet
1000 classes
~10 000 images per classes
1.4M in total
Top-1 ou Top-5 accuracy
AlexNet
First deep networks to win "ImageNet competition"
Krizhevsky, 2012
VGG
Even deeper, more regular kernels (3x3)
Simonyan, 2014
GoogLeNet
"We need to go deeper"
Szegedy, 2014
Depth is not everything
Or is it? : ResNet
He, 2015
SOTA
Xie, 11 Nov 2019
Transfer Learning
Strategies
When?
Generative Adversarial Networks
Models
Value function
Goodfellow, 2014
Conditioning
Mirza, 2014
Unsupervised Image to Image Translation
Zhu, 2017
Merci!
CNNs @ 3IT Victor Schmidt & Simon Verret Mila 3IT - Université de Sherbrooke 14 Novembre 2019