Python implementation of the model and learning algorithm proposed by Shim et al., 2018
This is a deep learning method to solve multi-class classification problems in a cost-sensitive setting. Given dataset whose feature acquisition costs are defined, our model tries to find the optimal feature acquisition policy and the classifier working with the policy. After training is done, this code prints out the test results (classification performance, # of times that each feature is acquired, etc.) of training/validation/test dataset.
CUBE dataset (used in the paper) is generated as default. You can also use your own dataset in the form of
the csv file whose first column is label followed by features values. Set
data_type
argument as 'csv' and pass csv filename as a keyword argument named
csv_filename
to data_load
function. You should define
feature acquisition cost and pass it to r_cost
argument.
1. Fork & Clone : Fork this project to your repository and clone to your work directory.
$ git clone https://github.com/OpenXAIProject/Joint-AFA-Classification.git
2. Run : Run python3 main.py --data_type=csv or cube_[feature dimension]_[sigma]. You can check additional options by following command.
$ python3 main.py --help
- python 3.5
- pytorch (0.4.1)
- numpy (1.15.0)
- matplotlib (2.2.2)
- scikit-learn (0.19.1)
If you found the provided code useful, please cite our work.
@inproceedings{shim2018jointAFA,
author = {Hajin Shim and Sung Ju Hwangand Eunho Yang and },
title = {Joint Active Feature Acquisition and Classification with Variable-Size Set Encoding},
booktitle = {NIPS},
year = {2018}
}
If you have any question, please contact Hajin Shim([email protected]).
This work was supported by Institute for Information & Communications Technology Promotion(IITP) grant funded by the Korea government(MSIT) (No.2017-0-01779, A machine learning and statistical inference framework for explainable artificial intelligence)
-
Project Name : A machine learning and statistical inference framework for explainable artificial intelligence(의사결정 이유를 설명할 수 있는 인간 수준의 학습·추론 프레임워크 개발)
-
Participated Affiliation : UNIST, Korea Univ., Yonsei Univ., KAIST, AItrics
-
Web Site : http://openXai.org