Summary of Gnns-to-mlps by Teacher Injection and Dirichlet Energy Distillation, By Ziang Zhou et al.
GNNs-to-MLPs by Teacher Injection and Dirichlet Energy Distillation
by Ziang Zhou, Zhihao Ding, Jieming Shi, Qing Li, Shiqi Shen
First submitted to arxiv on: 15 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Social and Information Networks (cs.SI)
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 proposes a novel method, called TINED, to distill Graph Neural Networks (GNNs) into multi-layer perceptrons (MLPs) for faster inference. By layer-wise distillation, TINED leverages the fine-grained knowledge within GNN layers and injects valuable teacher parameters from an FT in a GNN into an FC layer of the student MLP. The approach utilizes Dirichlet Energy Distillation to quantify the smoothing effects of FT and GP operations in GNN layers and distill these characteristics to MLP layers. Experimental results demonstrate that TINED achieves superior performance over GNNs and state-of-the-art distillation methods on seven datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary TINED is a new way to make Graph Neural Networks (GNNs) run faster. Currently, GNNs are great at classifying nodes in graphs, but they can be slow because they need to look at many layers of information during inference. TINED takes the good parts from GNNs and copies them into multi-layer perceptrons (MLPs), which can work much faster. The authors of this paper figured out how to do this layer by layer, using special techniques called Teacher Injection and Dirichlet Energy Distillation. They tested TINED on seven different datasets and found that it performed better than GNNs and other methods. |
Keywords
* Artificial intelligence * Distillation * Gnn * Inference
