The workshop is a slightly advanced practical introduction to graph neural networks in Python. We’ll start with a theoretical introduction, then we’ll move on to practical part. The practical part will consist of 3 main sub-parts:

• Preparing your own datasets

• Model implementations using Spektral and functional TensorFlow / Keras API

• Model implementations using Spektral and TensorFlow / Keras model sub-classing

Brief Bullet Point Outline:

• Introduction & theory (15 min)

• Spektral: layers, datasets and data loaders (15 min)

• Practice: Node-level classificatrion (20 min)

• Practice: Graph-level classification (20 min)

• Building your own dataset (10 min)

• Q&A (10 min)

Prerequisites:

People of all backgrounds and experience levels are welcome to the workshop. However, to get the most out of the workshop, the following skills are recommended:

• Basic understanding of graph structures

• Good understanding of basic machine learning & deep learning concepts

• Good understanding of Python

• Good understanding of Keras workflows (incl. functional API and model sub-classing)

• Basic understanding of TensorFlow 2.x

Aleksander Molak

Affiliation: Tensorcell

Currently engaged in a non-profit research initiative at Tensorcell, where I work with graph neural networks. For the last 3.5 years, I worked as an Innovation Lead and Machine Learning Researcher at Data Science and Artificial Intelligence Center of Excellence at Lingaro, building end-to-end machine learning systems for our global customers.

I am also a speaker and a blogger, author of #SundayAiPapers - a weekly LinkedIn microblog presenting the most recent papers on natural language processing, causal inference and probabilistic modeling and Medium blogs author. I am interested in NLP, causality, probabilistic modeling and graph neural networks.

I love traveling with my wife. I am passionate about vegan food, languages and running.

visit the speaker at: Github • Homepage