AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Compositional Fairness Constraints for Graph Embeddings

Avishek Bose, William Hamilton

ICML 2019 · Graph learning › Graph self-supervised learning · approach: Heterophily and imbalance on graphs · 116 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add compositional adversarial filters for gender, age, and occupationMovieLens1M embeddings · MovieLens1MRMSE: 0.865 -> 1.01Hurt
Add compositional adversarial filters for sensitive attributesMovieLens1M embeddings · MovieLens1MGender AUC for sensitive-attribute prediction: 0.712 -> 0.511Improved
Add compositional adversarial filters for sensitive attributesFreebase15k-237 TransD · Freebase15k-237Mean rank: 285 -> 542Hurt
Add compositional adversarial filters rather than separate per-attribute adversarial modelsMovieLens1M embeddings · MovieLens1MGender AUC for sensitive-attribute prediction: 0.532 -> 0.511Improved
Add compositional adversarial filters rather than separate per-attribute adversarial modelsFreebase15k-237 TransD · Freebase15k-237Mean rank: 320 -> 542Hurt
Increase adversarial regularization strengthCompositional graph embeddings · MovieLens1MRMSEHurt
Train with random combinations of sensitive-attribute filtersCompositional encoder · Reddit, held-out attribute combinationsSensitive-attribute AUC: 0.025Improved
Increase the number of sensitive attributes from 20 to 50Compositional adversary · Reddit, held-out combinationsHeld-out AUC: 0.569; 0.569; 0.556; 0.519Mixed
Use compositional rather than single-attribute adversarial trainingReddit graph embeddings · RedditSensitive-attribute predictionHurt
Use compositional rather than single-attribute adversarial trainingReddit graph embeddings · RedditEdge-prediction performanceHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy