TY - GEN
T1 - Learning a spatial ensemble of classifiers for raster classification
T2 - 14th IEEE International Conference on Data Mining Workshops, ICDMW 2014
AU - Jiang, Zhe
AU - Shekhar, Shashi
AU - Kamzin, Azamat
AU - Knight, Joseph
PY - 2015/1/26
Y1 - 2015/1/26
N2 - Given a spatial raster framework F, a set of explanatory feature maps, training and test samples with class labels on F, as well as a base classifier type, the problem of ensemble learning in raster classification aims to learn a collection of base classifiers to minimize classification errors. The problem has important societal applications such as land cover classification but is challenging due to existence of class ambiguity from spatial heterogeneity, i.e., Samples with the same feature values may have distinct class labels in different areas. Many existing approaches are non-spatial ensembles (e.g., Bagging, boosting, random forest), which assume that learning samples follow an identical distribution. Some spatial ensemble approaches also exist, which simply partition the raster framework into several regular sub-blocks and combine classification results on each sub-block. However, these existing approaches can not address the class ambiguity issue among pixels. In contrast, this paper proposes a new spatial ensemble approach, which partitions the spatial framework into several spatial footprints to minimize class ambiguity of training samples and then learns a base classifier for each footprint. Experimental evaluations on a real world remote sensing dataset show that the proposed spatial ensemble approach outperforms existing approaches when strong class ambiguity exists.
AB - Given a spatial raster framework F, a set of explanatory feature maps, training and test samples with class labels on F, as well as a base classifier type, the problem of ensemble learning in raster classification aims to learn a collection of base classifiers to minimize classification errors. The problem has important societal applications such as land cover classification but is challenging due to existence of class ambiguity from spatial heterogeneity, i.e., Samples with the same feature values may have distinct class labels in different areas. Many existing approaches are non-spatial ensembles (e.g., Bagging, boosting, random forest), which assume that learning samples follow an identical distribution. Some spatial ensemble approaches also exist, which simply partition the raster framework into several regular sub-blocks and combine classification results on each sub-block. However, these existing approaches can not address the class ambiguity issue among pixels. In contrast, this paper proposes a new spatial ensemble approach, which partitions the spatial framework into several spatial footprints to minimize class ambiguity of training samples and then learns a base classifier for each footprint. Experimental evaluations on a real world remote sensing dataset show that the proposed spatial ensemble approach outperforms existing approaches when strong class ambiguity exists.
KW - class ambiguity
KW - raster classification
KW - remote sensing
KW - spatial ensemble
KW - spatial heterogeneity
UR - http://www.scopus.com/inward/record.url?scp=84936850714&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84936850714&partnerID=8YFLogxK
U2 - 10.1109/ICDMW.2014.166
DO - 10.1109/ICDMW.2014.166
M3 - Conference contribution
AN - SCOPUS:84936850714
T3 - IEEE International Conference on Data Mining Workshops, ICDMW
SP - 15
EP - 18
BT - Proceedings - 14th IEEE International Conference on Data Mining Workshops, ICDMW 2014
A2 - Zhou, Zhi-Hua
A2 - Wang, Wei
A2 - Kumar, Ravi
A2 - Toivonen, Hannu
A2 - Pei, Jian
A2 - Zhexue Huang, Joshua
A2 - Wu, Xindong
PB - IEEE Computer Society
Y2 - 14 December 2014
ER -