CNN Architectures

Principali argomenti trattati:

• AlexNet
• VGG
• Residual Network
• Inception Network
• Data Augmentation
• Transfer Learning

La lezione è corredata di slides e di un notebook disponibile qui.

Importante Il secondo esonero riguarderà approfondimenti sulle reti convoluzionali. I dettagli dell’esonero sono pubblicati nell’apposita pagina.

Riferimenti bibliografici

• [Davies18], ch. 15.
• [Elg20], ch. 4.6, 5, 6.

Link utili

• Yann LeCun’s page on LeNet (with papers and description).

• ImageNet Classification with Deep Convolutional Neural Networks

• Local Response Norm vs. Bach Norm

• Very Deep Convolutional Networks for Large Scale Image Recognition

• Going Deeper with Convolutions

• A simple guide to the versions of Inception Networks

• Batch normalization: accelerating deep network training by reducing internal covariate shift

• Residual Learning for image Recognition

• How transferable are features in deep neural networks?

• Torch pretrained models

• TorchHUB

• Caltech101 dataset

• Survey on data augmentation technique