Summary of Graph Neural Network with Two Uplift Estimators For Label-scarcity Individual Uplift Modeling, by Dingyuan Zhu et al.
Graph Neural Network with Two Uplift Estimators for Label-Scarcity Individual Uplift Modeling
by Dingyuan Zhu, Daixin Wang, Zhiqiang Zhang, Kun Kuang, Yan Zhang, Yulin Kang, Jun Zhou
First submitted to arxiv on: 11 Mar 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 A novel graph neural network-based framework, called GNUM, is proposed for uplift modeling, which measures the incremental effect of a strategy or action on users from randomized experiments or observational data. Existing methods typically rely on individual data, which are limited in capturing unobserved factors affecting uplift. The scarcity of labeled data, especially for the treatment group, poses a significant challenge. GNUM leverages social graph features and relationships to estimate uplift, comprising two estimators: one generalizing all outcomes using class-transformed targets, and another utilizing partial labels from both treatment and control groups to alleviate label scarcity. Comprehensive experiments on public and industrial datasets demonstrate superior performance over state-of-the-art methods under various evaluation metrics. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Uplift modeling helps measure the effect of strategies or actions on people. Existing methods only use individual data, which can’t capture all the factors that affect this impact. It’s hard to find labeled data, especially for the group receiving the strategy. A new approach called GNUM uses social connections and relationships to estimate uplift. It has two parts: one that works with any outcome by transforming targets, and another that uses partial labels from both groups to handle limited labeled data. This approach performs better than others in real-world scenarios. |
Keywords
* Artificial intelligence * Graph neural network
