Context. A task of data classification under conditions of clusters’ overlapping is considered in this article. Besides that, it’s assumed<br />that information to be processed is given in the rank scale.<br />Objective. It’s proposed to use a double neo-fuzzy neuron for classification which is a modification of a traditional neo-fuzzy neuron<br />with specially designed asymmetrical membership functions and improved approximating properties.<br />Method. The double neo-fuzzy neuron (just like the traditional one) is designated for processing data given the scale of natural numbers.<br />However, the situation may become complicated greatly if source data is not given in the numerical scale but in the ordinal one which is a<br />quite common case for a wide variety of practical tasks.<br />Results. A gradient minimization procedure with a variable learning step parameter was used for learning the double neo-fuzzy neuron.<br />The proposed approach to fuzzy classification for data given in the ordinal scale based on the double neo-fuzzy neuron which is learnt with<br />the help of a high-speed algorithm possesses additional smoothing properties. The clustering accuracy for a training sample and the test one as well as the system’s learning speed were measured during experiments. The proposed architecture of the double neo-fuzzy neuron is a sort of compromise between a traditional neo-fuzzy neuron and its extended modification. This architecture demonstrates good performance in those cases when the results’ accuracy has more influence compared to the elapsed time used for data processing.<br />Conclusions. Experimental implementation (for both artificial and real-world data) proved efficiency of the proposed techniques.<br />During the experiments, properties of the proposed system were studied which confirmed usability of the proposed system for a wide range of Data Mining tasks.
Науковий журнал «Радіоелектроніка, інформатика, управління»
Переглянути архів Інформація| Поле | Співвідношення | |
| ##plugins.schemas.marc.fields.042.name## |
dc |
|
| ##plugins.schemas.marc.fields.720.name## |
Zhengbing, Hu; School of Educational Information Technology, Central China Normal University, Wuhan, China Bodyanskiy, Ye. V.; Kharkiv National University of Radio Electronics, Kharkiv, Ukraine Tyshchenko, O. K.; Kharkiv National University of Radio Electronics, Kharkiv, Ukraine Samitova, V. O.; Kharkiv National University of Radio Electronics, Kharkiv, Ukraine |
|
| ##plugins.schemas.marc.fields.520.name## |
Context. A task of data classification under conditions of clusters’ overlapping is considered in this article. Besides that, it’s assumed<br />that information to be processed is given in the rank scale.<br />Objective. It’s proposed to use a double neo-fuzzy neuron for classification which is a modification of a traditional neo-fuzzy neuron<br />with specially designed asymmetrical membership functions and improved approximating properties.<br />Method. The double neo-fuzzy neuron (just like the traditional one) is designated for processing data given the scale of natural numbers.<br />However, the situation may become complicated greatly if source data is not given in the numerical scale but in the ordinal one which is a<br />quite common case for a wide variety of practical tasks.<br />Results. A gradient minimization procedure with a variable learning step parameter was used for learning the double neo-fuzzy neuron.<br />The proposed approach to fuzzy classification for data given in the ordinal scale based on the double neo-fuzzy neuron which is learnt with<br />the help of a high-speed algorithm possesses additional smoothing properties. The clustering accuracy for a training sample and the test one as well as the system’s learning speed were measured during experiments. The proposed architecture of the double neo-fuzzy neuron is a sort of compromise between a traditional neo-fuzzy neuron and its extended modification. This architecture demonstrates good performance in those cases when the results’ accuracy has more influence compared to the elapsed time used for data processing.<br />Conclusions. Experimental implementation (for both artificial and real-world data) proved efficiency of the proposed techniques.<br />During the experiments, properties of the proposed system were studied which confirmed usability of the proposed system for a wide range of Data Mining tasks. |
|
| ##plugins.schemas.marc.fields.260.name## |
Zaporizhzhya National Technical University 2017-05-13 11:57:04 |
|
| ##plugins.schemas.marc.fields.856.name## |
application/pdf http://ric.zntu.edu.ua/article/view/101031 |
|
| ##plugins.schemas.marc.fields.786.name## |
Radio Electronics, Computer Science, Control; No 1 (2017): Radio Electronics, Computer Science, Control |
|
| ##plugins.schemas.marc.fields.540.name## |
Copyright (c) 2017 Hu Zhengbing, Ye. V. Bodyanskiy, O. K. Tyshchenko, V. O. Samitova |
|