/SpectralRankingWithCovariates

Code for the research project -- Spectral Ranking with Covariates

Primary LanguagePython

Spectral Ranking with Covariates

This repository contains the code for the project Spectral Ranking with Covariates.

Algorithms and where to find them

You will find implementations of the following ranking algorithms from the following scripts:

src/spektrankers.py
# Our proposed methods
├── SVDRankerNormal (SVDRank)
├── SVDRankerCov (SVDCovRank)
├── SVDRankerKCov (Kernelised SVDCovRank)
├── SerialRank (SerialRank)
├── CSerialRank (C-SerialRank)
├── CCARank (CCRank)
├── KCCARank (Kernelised CCRank)
# Spectral ranking benchmarks
├── RankCentrality (Rank Centrality) 
└── DiffusionRankCentrality (Regularised Rank Centrality)

# Probabilistic ranking benchmarks
src/prefkrr.py
└── PreferentialKRR (Bradley Terry with GP link)
src/baselines.py
├── BradleyTerryRanker (Bradley Terry Model)
└── Pairwise_LogisticRegression (Bradley Terry with logistic regressoin)

The algorithms used in this repo came primarily out of the following papers. If you use them in your research we would appreciate a citation to our paper:

@inproceedings{chau2022spectral,
  title={Spectral ranking with covariates},
  author={Chau, Siu Lun and Cucuringu, Mihai and Sejdinovic, Dino},
  booktitle={Joint European Conference on Machine Learning and Knowledge Discovery in Databases},
  pages={70--86},
  year={2022},
  organization={Springer}
}