Graph-Based Machine Learning with Python
summary
Covers how graph ML makes node and graph predictions using actual Python code and numbers: uses NetworkX for analysis and PyTorch Geometric to build a two‑layer GCN with message‑passing, applied to the Cora citation network (≈2,700 papers, ≈5,400 edges, label set corrected to 70 topics). Claims: GCN trains in under 30 seconds vs a gradient‑boosting baseline that took ~4 hours and performs substantially better by leveraging citation links. Notes real uses (fraud detection, recommender systems, molecular tasks), shallow GNNs, and scaling concerns for billion‑node graphs.
AI-written summary · about
chapters (9)
- 0:00Introduction and speaker background
- 2:53What are graphs and their uses
- 6:25Graphs in machine learning
- 8:14NetworkX library overview
- 11:45Graph neural networks explained
- 15:55PyTorch Geometric implementation
- 18:03Real-world example with Cora dataset
- 21:11Advantages of graph neural networks
- 22:16Q&A session
related talks
47mPyTorch Geometric for Graph Neural Nets
23mGraph ML – The Next Level Of Machine Learning; Learn How To Use It And When Not To
29mGraph Neural Networks + LLM on
39mUsing Graph Neural Networks to Embrace the Dependency Within Your Data
01h 06mMachine Learning over Graphs
40mRun Large-Scale Graph Analytics Using The Most Popular Graph Analytics Library Available
59mTransforming DataFrames: Intro to Graphs & GNNs with PyTorch Geometric
01h 30mPractical graph neural networks in Python with TensorFlow and Spektral