Summary of Replay-and-forget-free Graph Class-incremental Learning: a Task Profiling and Prompting Approach, by Chaoxi Niu et al.
Replay-and-Forget-Free Graph Class-Incremental Learning: A Task Profiling and Prompting Approach
by Chaoxi Niu, Guansong Pang, Ling Chen, Bing Liu
First submitted to arxiv on: 14 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 The paper presents a novel approach to class-incremental learning (CIL) on graph data, where the goal is to learn a sequence of tasks with unique classes. The key challenge lies in separating classes from different tasks without knowing task identifiers during inference. To address this issue, the authors propose a Laplacian smoothing-based graph task profiling approach that models each graph task by a task prototype. This approach guarantees distinct task prototypes for different tasks while being nearly identical for tasks of the same class. Additionally, the authors introduce a novel graph prompting approach for GCIL that learns a small discriminative graph prompt for each task, resulting in a separate classification model for each task. The prompt learning requires training a single graph neural network (GNN) only once on the first task, and no data replay is required thereafter. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to learn a sequence of tasks with unique classes on graph data. Without knowing which task you’re working on, it’s hard to tell what class something belongs to. The authors create a special profile for each task that captures its unique characteristics. This helps them separate classes from different tasks correctly. They also develop a technique called graph prompting that learns a small prompt for each task, allowing the model to work well even when it forgets some of the knowledge it learned earlier. |
Keywords
» Artificial intelligence » Classification » Gnn » Graph neural network » Inference » Prompt » Prompting
