Monday, March 18, 2024 - 14:00
The School of Computer Science is pleased to present...
Technical Workshop Series
ML, NN, and GNN using PyTorch and PyG – First Session: PyTorch
Presenter: Soroush Ziaeinejad
Date: Monday, March 18th, 2024
Time: 2:00 pm
Location: 4th Floor (Workshop space) at 300 Ouellette Avenue (School of Computer Science Advanced Computing Hub)
Abstract:
This workshop provides an overview of a significant framework in deep learning: PyTorch (and PyTorch Geometric.) The first session covers PyTorch, addressing its core concepts, architecture design, and model optimization processes. The second session introduces PyTorch Geometric, which focuses on applying graph neural networks to analyze complex graph data. Participants will engage with practical examples and theoretical discussions, enhancing their understanding of these frameworks' roles in advancing deep learning research. This workshop is aimed at researchers and students seeking to deepen their knowledge of deep learning technologies.
Workshop Outline:
-
Introduction to PyTorch: Understanding the core concepts and its dynamic computation graph.
-
Tensors and Operations: Basics of tensor operations are fundamental to neural networks.
-
Building Blocks of Neural Networks: Detailed exploration of layers, activation, and loss functions.
-
Training Neural Networks: A step-by-step guide to training models, including backpropagation, optimization, and regularization techniques.
-
Practical Application: Implement a simple neural network to solve a classification or regression problem, illustrating the workflow from data preparation to model evaluation.
Prerequisites:
-
Basic programming experience in Python.
-
Understanding of fundamental machine learning concepts.
-
Familiarity with the core principles of neural networks is helpful but not mandatory.
Biography:
Soroush is a Ph.D. student and research assistant at the School of Computer Science. His main research area is Natural Language Processing and Information Retrieval on social networks.