Skip to content

git-disl/HQ-Ensemble

Repository files navigation

HQ-Ensemble: Hierarchical Ensemble Pruning


GitHub license Version

Introduction

If you find this work useful in your research, please cite the following paper:

Bibtex:

@InProceedings{hq-ensemble,
author={Wu, Yanzhao and Liu, Ling},
booktitle={2021 IEEE International Conference on Data Mining (ICDM)}, 
title={{Boosting Deep Ensemble Performance with Hierarchical Pruning}}, 
volume={},
number={},
pages={1433-1438},
month = {Dec.},
year = {2021}
}

Instructions

Following the steps below for using our HQ-Ensemble for efficient ensemble pruning for a dataset <dataset>.

  1. Install then dependencies in requirements.txt, and obtain the pretrained models for the dataset <dataset> according to the model files under <dataset> folder.
  2. Extract the prediction vectors and labels for <dataset> and store them as numpy vectors under <dataset>/prediction for testing data and <dataset>/train for training data.
  3. Set up the environments with env.sh, then execute the HQ-Ensemble.py or baselineDiversityPruning.py file to obtain the corresponding results.

Please refer to our paper and appendix for detailed results.

Problem

Installation

pip install -r requirements.txt

Supported Platforms

Development / Contributing

Issues

Status

Contributors

See the people page for the full listing of contributors.

License

Copyright (c) 20XX-20XX Georgia Tech DiSL
Licensed under the Apache License.