source code for CDR-FRDpA paper
CDR-FRDpA enhances the FCM-RDpA (Fuzzy C-Means Clustering, Regularization, DropRule, and Powerball AdaBelief; paper|arxiv|code|blog) via consistent dimensionality reduction to optimize TSK fuzzy systems for regression in high dimensionality
run demoCDR.m to reproduce the results on the Estate-costs dataset of Fig.2/3 in the paper.
run demoPS.m to reproduce the results on the Estate-costs dataset of Fig.4 in the paper.
run demoInit.m to reproduce the results on the Estate-costs dataset of Fig.5 in the paper.
run demoMF.m to reproduce the results on the Estate-costs dataset of Fig.7 in the paper.
run demoRP.m to reproduce the results on the Estate-costs dataset of Fig.8 in the paper.
We also provide a sugfis_mbgd_app.mlapp for simple test. Some examples are given as below:
@Article{Shi2021,
author = {Zhenhua Shi and Dongrui Wu and Chenfeng Guo and Changming Zhao and Yuqi Cui and Fei-Yue Wang},
journal = {Information Sciences},
title = {{FCM-RDpA}: {TSK} Fuzzy Regression Model Construction Using Fuzzy C-Means Clustering, Regularization, {D}rop{R}ule, and {P}owerball {A}da{B}elief},
year = {2021},
pages = {490-504},
volume = {574},
}
@Article{Wu2020,
author = {Dongrui Wu and Ye Yuan and Jian Huang and Yihua Tan},
journal = {IEEE Trans. on Fuzzy Systems},
title = {Optimize {TSK} Fuzzy Systems for Regression Problems: Mini-batch Gradient Descent With Regularization, {D}rop{R}ule, and {A}da{B}ound ({MBGD-RDA})},
year = {2020},
number = {5},
pages = {1003-1015},
volume = {28},
}