Summary of Derivation Of Back-propagation For Graph Convolutional Networks Using Matrix Calculus and Its Application to Explainable Artificial Intelligence, by Yen-che Hsiao et al.
Derivation of Back-propagation for Graph Convolutional Networks using Matrix Calculus and its Application to Explainable Artificial Intelligence
by Yen-Che Hsiao, Rongting Yue, Abhishek Dutta
First submitted to arxiv on: 2 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents a detailed derivation of the backpropagation algorithm for graph convolutional neural networks (GCNNs) using matrix calculus. The authors extend this derivation to include arbitrary element-wise activation functions and an arbitrary number of layers, enabling GCNNs to tackle node classification and link prediction tasks. To validate their method, they compare it with reverse-mode automatic differentiation, demonstrating median sum of squared errors within the range of 10^(-18) to 10^(-14). Experimental results are presented on a five-layer GCNN applied to Zachary’s karate club social network for node classification and a drug-drug interaction network for link prediction. The derived closed-form solution has implications for explainable AI and sensitivity analysis. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows how to make computers learn from complex data, like who is friends with whom in a social network or which drugs interact with each other. They use special math to make a new type of computer program that can solve these kinds of problems better than before. The program is tested on two different networks and does very well. This means we might be able to use this kind of program to understand how computers are making decisions, and why they’re getting things right or wrong. |
Keywords
* Artificial intelligence * Backpropagation * Classification
