HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification
To overcome the high intramodel dimensionality and low ensemble diversity issues, which limit the classification performance of original deep forest (DF), a new version of DF, the high-ordinary least square projection (HOLP) DF, was proposed in this article by introducing model-based HOLP feature sc...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9899729/ |
| _version_ | 1829097117133570048 |
|---|---|
| author | Alim Samat Erzhu Li Wei Wang Sicong Liu Ximing Liu |
| author_facet | Alim Samat Erzhu Li Wei Wang Sicong Liu Ximing Liu |
| author_sort | Alim Samat |
| collection | DOAJ |
| description | To overcome the high intramodel dimensionality and low ensemble diversity issues, which limit the classification performance of original deep forest (DF), a new version of DF, the high-ordinary least square projection (HOLP) DF, was proposed in this article by introducing model-based HOLP feature screening (FS), random subspace propagation, and reduced error pruning techniques. To evaluate the performance of the proposed HOLP-DF, total eleven popular FS algorithms and total six advanced deep learning methods are selected. Experimental results on three widely acknowledged hyperspectral and PolSAR image classification benchmarks showed that: 1) HOLP is an optimal choice for FS in contrast with other screeners in terms of high classification accuracy and execution efficiency; 2) HOLP-DF is capable of obtaining better results than the original DF, DF with confidence screening and feature screening; 3) optimum sets of model depth, propaganda ratio and screening ratio parameters are 30, 40%, and 40%, respectively; 4) performance of HOLP-DF can be further boosted by extra usage of patch-based pooling and morphological profiling techniques. |
| first_indexed | 2024-04-11T09:08:24Z |
| format | Article |
| id | doaj.art-7fbc8253650f44b68197e3a1fd590b64 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-04-11T09:08:24Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-7fbc8253650f44b68197e3a1fd590b642022-12-22T04:32:34ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352022-01-01158287829810.1109/JSTARS.2022.32068869899729HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image ClassificationAlim Samat0https://orcid.org/0000-0002-9091-6033Erzhu Li1https://orcid.org/0000-0002-5881-618XWei Wang2https://orcid.org/0000-0003-1813-0551Sicong Liu3https://orcid.org/0000-0003-1612-4844Ximing Liu4State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, ChinaDepartment of Geographical Information Science, Jiangsu Normal University, Xuzhou, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, ChinaCollege of Surveying and Geoinformatics, Tongji University, Shanghai, ChinaState Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, ChinaTo overcome the high intramodel dimensionality and low ensemble diversity issues, which limit the classification performance of original deep forest (DF), a new version of DF, the high-ordinary least square projection (HOLP) DF, was proposed in this article by introducing model-based HOLP feature screening (FS), random subspace propagation, and reduced error pruning techniques. To evaluate the performance of the proposed HOLP-DF, total eleven popular FS algorithms and total six advanced deep learning methods are selected. Experimental results on three widely acknowledged hyperspectral and PolSAR image classification benchmarks showed that: 1) HOLP is an optimal choice for FS in contrast with other screeners in terms of high classification accuracy and execution efficiency; 2) HOLP-DF is capable of obtaining better results than the original DF, DF with confidence screening and feature screening; 3) optimum sets of model depth, propaganda ratio and screening ratio parameters are 30, 40%, and 40%, respectively; 4) performance of HOLP-DF can be further boosted by extra usage of patch-based pooling and morphological profiling techniques.https://ieeexplore.ieee.org/document/9899729/Deep forest (DF)feature screeninghigh-ordinary least square projection (HOLP)hyperspectralimage classificationPolSAR |
| spellingShingle | Alim Samat Erzhu Li Wei Wang Sicong Liu Ximing Liu HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Deep forest (DF) feature screening high-ordinary least square projection (HOLP) hyperspectral image classification PolSAR |
| title | HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification |
| title_full | HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification |
| title_fullStr | HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification |
| title_full_unstemmed | HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification |
| title_short | HOLP-DF: HOLP Based Screening Ultrahigh Dimensional Subfeatures in Deep Forest for Remote Sensing Image Classification |
| title_sort | holp df holp based screening ultrahigh dimensional subfeatures in deep forest for remote sensing image classification |
| topic | Deep forest (DF) feature screening high-ordinary least square projection (HOLP) hyperspectral image classification PolSAR |
| url | https://ieeexplore.ieee.org/document/9899729/ |
| work_keys_str_mv | AT alimsamat holpdfholpbasedscreeningultrahighdimensionalsubfeaturesindeepforestforremotesensingimageclassification AT erzhuli holpdfholpbasedscreeningultrahighdimensionalsubfeaturesindeepforestforremotesensingimageclassification AT weiwang holpdfholpbasedscreeningultrahighdimensionalsubfeaturesindeepforestforremotesensingimageclassification AT sicongliu holpdfholpbasedscreeningultrahighdimensionalsubfeaturesindeepforestforremotesensingimageclassification AT ximingliu holpdfholpbasedscreeningultrahighdimensionalsubfeaturesindeepforestforremotesensingimageclassification |