GraphPowered Machine Learning First Steps
From graph analytics to graph neural networks: Making the most of your graph data
Topic: Data
Many powerful machine learning algorithms—including PageRank (Pregel), recommendation engines (collaborative filtering), and text summarization and other NLP tasks—are based on graphs. And there are even more applications once you consider data preprocessing and feature engineering, which are both vital tasks in machine learning pipelines.
Join expert Jörg Schad to explore the symbiosis of graphs and machine learning, starting with graph analytics to graph neural networks. You’ll learn why graphs are such a powerful abstraction and discover how to leverage them in your machine learning projects.
What you'll learnand how you can apply it
By the end of this live online course, you’ll understand:
 Why graphs are such a powerful abstraction
 How to leverage knowledge graphs
 The graph ecosystem including its many many powerful open source tools
 How to extract value from graphs using graph analytics and graph algorithms
 How to combine deep learning and graphs
 How we can learn graphs features using graph neural networks
And you’ll be able to:
 Identify suitable graph use cases
 Use open source tools to create, manipulate, and query graphs
 Leverage graph algorithms to analyze graphs
 Apply machine learning techniques on graphs
 Utilize the latest graph neural network techniques
This training course is for you because...
 You’re a data scientist who wants to extend your toolset.
 You’re a technical leader or architect who wants to understand the power of graphs and machine learning.
 You’re an engineer who wants to learn about graphs and machine learning.
Prerequisites
 Familiarity with machine learning concepts and tools (e.g., TensorFlow) and Python (useful but not required)
Recommended preparation:
 Read “Preliminaries,” “Python Language Basics, IPython, and Jupyter Notebooks,” and “BuiltIn Data Structures, Functions, and Files” (chapters 1–3 in Python for Data Analysis, second edition)
Recommended followup:
 Finish Python for Data Analysis, second edition (book)
About your instructor

Jörg Schad is Head of Machine Learning at ArangoDB. In a previous life, he has worked on or built machine learning pipelines in healthcare, distributed systems at Mesosphere, and inmemory databases. He received his Ph.D. for research around distributed databases and data analytics. He’s a frequent speaker at meetups, international conferences, and lecture halls.
Schedule
The timeframes are only estimates and may vary according to how the class is progressing
Course overview (15 minutes)
 Presentation: Jupyter Notebook and infrastructure overview
Graph basics and queries (40 minutes)
 Presentation: What are graphs, and why are they so powerful?; graph properties and characteristics; graph storage and queries
 Handson exercises: Work with SPARQL; work with property graphs
 Q&A
 Break (5 minutes)
Graph analytics and algorithms (40 minutes)
 Presentation: Graph analytics and algorithms; identifying graph problems; complex graphs use cases
 Handson exercises: Use graph algorithms for reachability and shortest path queries; implement basic graph analytics—collaborative filtering; use PageRank; explore graphbased fraud detection
 Q&A
 Break (5 minutes)
Graph toolset (40 minutes)
 Presentation: Open source graph tools; graph embeddings for machine learning
 Handson exercise: Use NetworkX; use the Deep Graph Library; generate graph embeddings with node2vec
 Q&A
 Break (5 minutes)
Graph neural network (40 minutes)
 Presentation: Introduction to graph neural networks; advanced graph neural networks
 Handson exercises: Build a graph convolutional network from scratch; use the Deep Graph Library and advanced graph convolutional networks
 Q&A
 Break (5 minutes)
Endtoend project (40 minutes)
 Presentation: The machine learning cycle; next steps
 Handson exercises: Explore citation graph data and predict data quality, from data preparation, model building, evaluation, and retraining to inference
Wrapup and Q&A (5 minutes)