http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Sparse pseudoinverse incremental extreme learning machine
Kassani, Peyman Hosseinzadeh,Teoh, Andrew Beng Jin,Kim, Euntai Elsevier 2018 Neurocomputing Vol.287 No.-
<P>An extreme learning machine (ELM) is a popular analytic single hidden layer feedforward neural network because of its rapid learning capacity. However, vanilla dense ELMs are affected by the overfitting problem when the number of hidden neurons is high. Further direct consequences of the density are decreases in both the training and prediction speeds. In this study, we propose an incremental method for sparsifying the ELM using a newly devised indicator driven by the condition number in the ELM design matrix, in which we call sparse pseudoinverse incremental-ELM (SPI-ELM). SPI-ELM exhibits better generalization performance and lower run-time complexity compared with ELM. However, the sparsification process may negatively affect the learning speed of SPI-ELM; thus, we introduce an iterative matrix decomposition algorithm to address this issue. We also demonstrate that there is a useful relationship between the condition number in the ELM design matrix and the number of hidden neurons. This relationship helps to understand the random weights and nonlinear activation functions in ELMs. We evaluated the SPI-ELM method based on 20 benchmark data sets from the University of California Irvine repository and three real-world databases from the computer vision domain. (C) 2018 Elsevier B.V. All rights reserved.</P>
Peyman Hosseinzadeh Kassani,Euntai Kim 한국지능시스템학회 2016 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.16 No.2
The proposal of this study is a fast version of the conventional extreme learning machine (ELM), called pseudoinverse matrix decomposition based incremental ELM (PDI-ELM). One of the main problems in ELM is to determine the number of hidden nodes. In this study, the number of hidden nodes is automatically determined. The proposed model is an incremental version of ELM which adds neurons with the goal of minimization the error of the ELM network. To speed up the model the information of pseudoinverse from previous step is taken into account in the current iteration. To show the ability of the PDI-ELM, it is applied to few benchmark classification datasets in the University of California Irvine (UCI) repository. Compared to ELM learner and two other versions of incremental ELM, the proposed PDI-ELM is faster.
Kassani, Peyman Hosseinzadeh,Kim, Euntai Korean Institute of Intelligent Systems 2016 INTERNATIONAL JOURNAL of FUZZY LOGIC and INTELLIGE Vol.16 No.2
The proposal of this study is a fast version of the conventional extreme learning machine (ELM), called pseudoinverse matrix decomposition based incremental ELM (PDI-ELM). One of the main problems in ELM is to determine the number of hidden nodes. In this study, the number of hidden nodes is automatically determined. The proposed model is an incremental version of ELM which adds neurons with the goal of minimization the error of the ELM network. To speed up the model the information of pseudoinverse from previous step is taken into account in the current iteration. To show the ability of the PDI-ELM, it is applied to few benchmark classification datasets in the University of California Irvine (UCI) repository. Compared to ELM learner and two other versions of incremental ELM, the proposed PDI-ELM is faster.
Proposing a GPU based Modified Fuzzy Nearest Neighbor Rule for Traffic Sign Detection
Peyman Hosseinzadeh Kassani,Junhyuk Hyun,Euntai Kim 제어로봇시스템학회 2015 제어로봇시스템학회 국제학술대회 논문집 Vol.2015 No.10
The purpose of this study is introducing a graphical process unit (GPU) implementation of a modified fuzzy nearest neighbor rule useful for traffic sign detection (TSD). The new method tries to detect road signs using color information in order to locate regions of interest. The candidate regions of interest are obtained by color information. Afterward, candidate regions are used for making histogram of oriented gradient (HOG) feature. Finally, the features are fed into the GPU-based modified fuzzy nearest neighbor in order to detect traffic signs. The proposed rule modifies the way for fuzzification of query sample in terms of distances while the conventional fuzzy nearest neighbor (FNN) doesn’t care distance of local neighbors. The accuracy of the proposed method is compared with the state of the arts k-nearest neighbor (k-NN), FNN and support vector machine algorithms on the challenging German traffic sign detection benchmark (GTSDB) data set. Results indicate that the modified rule achieves good accuracy and is competitive compared to others.