Analysis Algorithm Kohonen and Momentum on the Back Propagation Neural Network
Purwa Hasan Putra1, Muhammad Zarlis2, H Mawengkang3
1Purwa Hasan Putra, Graduate, School of Computer Science, Universitas Sumatera Utara, Malaysia.
2Muhammad Zarlis, Graduate, School of Computer Science, Universitas Sumatera Utara, Malaysia.
3H Mawengkang, Graduate, School of Mathematics, Universitas Sumatera Utara, Malaysia.
Manuscript received on 09 May 2019 | Revised Manuscript received on 19 May 2019 | Manuscript Published on 23 May 2019 | PP: 1349-1355 | Volume-7 Issue-6S5 April 2019 | Retrieval Number: F12340476S519/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In this paper, it is faster to achieve the target error with the weight addition of Kohonen and Momentum combination in Back propagation when compared with the addition of random weight. The error quadratic decrease in Back propagation method training with the target error of 0.007 with random weight reaches the target error at 55 iterations, whereas the weight addition of Kohonen and Momentum in Back propagation reaches the target error at 36 iterations. The test results with the weight addition of Kohonen and Momentum combination in Back propagation is better when compared with the addition of random weight, where it can recognize the test data reaching the accuracy of 96.53%.
Keywords: Neural Network, Kohonen, Momentum, Back Propagation.
Scope of the Article: Network Security Trust, & Privacy