Neural Networks

Grafi di computazione e modelli non lineari. Reti Convoluzionali. LeNet-5.

Materiale didattico:

• Modelli non lineari (slides)
• Reti convoluzionali (slides)
• Architetture convoluzionali
• Notebook di accompagnamento

Riferimenti bibliografici

• [Davies18], Ch. 15.
• [Gon18], Ch. 13.
• [Elg20], Ch. 2,3
• [Sze22], Ch. 5.3, 5.4, 6.2

Link utili

• Introduction to Neural Networks

• Dive into Deep Learning

• Course Notes on Optimization for Machine Learning and Mathematical Foundations of Data Science

• Understanding graphs and automatic differentiation

• Pytorch autograd: Understanding the heart of Pytorch’s magic

• Yes you should understand backpropagation

• Visualizing and understanding convolutional Networks

• Vincent Dumoulin, Francesco Visin - A guide to convolution arithmetic for deep learning]

• An illustrated explanation of performing 2D convolution using matrix multiplication. Un esempio illustrativo anche su slideshare.

• Visualizing neural networks using saliency maps. Reference paper su arxiv.

• Saliency maps in Flashtorch. Medium description.

• A collection of Deep network visualization techniques in Pytorch