Groupe d’études et de recherche en analyse des décisions

PyTorch tutorial for deep learning

Antoine Prouvost Polytechnique Montréal, Canada

Put your understanding of deep learning into practice with (one of) the most flexible deep learning framework.

The tutorial will show you how to implement a neural network along with some theoretical reminders, practical tips, and suggested code workflows. More precisely, we will explore how PyTorch can be used as a simple version of numpy, with GPU acceleration and automatic gradients computations. Afterwards, will explore how PyTorch can be used to train a neural network in an efficient way, with minimum code overhead. We'll used it to train a convolutional model for digit recognition.

Keywords: Tensors, Python, GPU, Automatic Differentiation, Neural Networks, Dynamic

Prerequisites:
● Python, numpy
● Machine learning basics: loss function, generalization
● Neural networks understanding: feed forward, backpropagation, gradient descent

Required: A computer (GPU isn't required) and a Google account

The tutorial will last until 17:00 PM for those who wishes to stay.

Registration at: https://goo.gl/forms/sb33ufXIVkIgCJ3e2