Ordinal classification is a class of special tasks in machine learning and pattern recognition. As to ordinal classification, there is an ordinal structure among different decision values. The monotonicity constraint between features and decision should be taken into account as the fundamental assumption. However, in real-world applications, this assumption may be not true. Only some candidate features, instead of all, are monotonic with decision. So the existing feature selection algorithms which are designed for nominal classification or monotonic classification are not suitable for ordinal classification. In this paper, we propose a feature selection algorithm for ordinal classification based on considering the non-monotonic and monotonic features separately. We first introduce an assumption of hybrid monotonic classification consistency and define a feature evaluation function to calculate the relevance between the features and decision for ordinal classification. Then, we combine the reported measure and genetic algorithm (GA) to search the optimal feature subset. A collection of numerical experiments are implemented to show that the proposed approach can effectively reduce the feature size and improve the classification performance.
Weiwei PAN
Xiamen University of Technology
Qinhua HU
Tianjin University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Weiwei PAN, Qinhua HU, "An Improved Feature Selection Algorithm for Ordinal Classification" in IEICE TRANSACTIONS on Fundamentals,
vol. E99-A, no. 12, pp. 2266-2274, December 2016, doi: 10.1587/transfun.E99.A.2266.
Abstract: Ordinal classification is a class of special tasks in machine learning and pattern recognition. As to ordinal classification, there is an ordinal structure among different decision values. The monotonicity constraint between features and decision should be taken into account as the fundamental assumption. However, in real-world applications, this assumption may be not true. Only some candidate features, instead of all, are monotonic with decision. So the existing feature selection algorithms which are designed for nominal classification or monotonic classification are not suitable for ordinal classification. In this paper, we propose a feature selection algorithm for ordinal classification based on considering the non-monotonic and monotonic features separately. We first introduce an assumption of hybrid monotonic classification consistency and define a feature evaluation function to calculate the relevance between the features and decision for ordinal classification. Then, we combine the reported measure and genetic algorithm (GA) to search the optimal feature subset. A collection of numerical experiments are implemented to show that the proposed approach can effectively reduce the feature size and improve the classification performance.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E99.A.2266/_p
Copy
@ARTICLE{e99-a_12_2266,
author={Weiwei PAN, Qinhua HU, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={An Improved Feature Selection Algorithm for Ordinal Classification},
year={2016},
volume={E99-A},
number={12},
pages={2266-2274},
abstract={Ordinal classification is a class of special tasks in machine learning and pattern recognition. As to ordinal classification, there is an ordinal structure among different decision values. The monotonicity constraint between features and decision should be taken into account as the fundamental assumption. However, in real-world applications, this assumption may be not true. Only some candidate features, instead of all, are monotonic with decision. So the existing feature selection algorithms which are designed for nominal classification or monotonic classification are not suitable for ordinal classification. In this paper, we propose a feature selection algorithm for ordinal classification based on considering the non-monotonic and monotonic features separately. We first introduce an assumption of hybrid monotonic classification consistency and define a feature evaluation function to calculate the relevance between the features and decision for ordinal classification. Then, we combine the reported measure and genetic algorithm (GA) to search the optimal feature subset. A collection of numerical experiments are implemented to show that the proposed approach can effectively reduce the feature size and improve the classification performance.},
keywords={},
doi={10.1587/transfun.E99.A.2266},
ISSN={1745-1337},
month={December},}
Copy
TY - JOUR
TI - An Improved Feature Selection Algorithm for Ordinal Classification
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2266
EP - 2274
AU - Weiwei PAN
AU - Qinhua HU
PY - 2016
DO - 10.1587/transfun.E99.A.2266
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E99-A
IS - 12
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - December 2016
AB - Ordinal classification is a class of special tasks in machine learning and pattern recognition. As to ordinal classification, there is an ordinal structure among different decision values. The monotonicity constraint between features and decision should be taken into account as the fundamental assumption. However, in real-world applications, this assumption may be not true. Only some candidate features, instead of all, are monotonic with decision. So the existing feature selection algorithms which are designed for nominal classification or monotonic classification are not suitable for ordinal classification. In this paper, we propose a feature selection algorithm for ordinal classification based on considering the non-monotonic and monotonic features separately. We first introduce an assumption of hybrid monotonic classification consistency and define a feature evaluation function to calculate the relevance between the features and decision for ordinal classification. Then, we combine the reported measure and genetic algorithm (GA) to search the optimal feature subset. A collection of numerical experiments are implemented to show that the proposed approach can effectively reduce the feature size and improve the classification performance.
ER -