Perbandingan Kinerja Fungsi Tangen-Hyperbolic Dan Rectified Linear Unit Dalam Pengklasifikasian Citra Daun Anggur Berbasis Neural Network

DAGANG, Yohanes Don Bosco (2023) Perbandingan Kinerja Fungsi Tangen-Hyperbolic Dan Rectified Linear Unit Dalam Pengklasifikasian Citra Daun Anggur Berbasis Neural Network. Undergraduate thesis, Universitas Katolik Widya Mandira Kupang.

	Text ABSTRAK.pdf Download (778kB)
	Text BAB I.pdf Download (36kB)
	Text BAB II.pdf Restricted to Repository staff only Download (315kB)
	Text BAB III.pdf Restricted to Repository staff only Download (44kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (351kB)
	Text BAB V.pdf Restricted to Repository staff only Download (990kB)
	Text BAB VI.pdf Download (279kB)

Abstract

As technology advances, all forms of human work are facilitated by the presence of new technologies that continue to develop. One of them is in the field of digital image processing. Image processing is very helpful in recognizing and classifying objects quickly and more processed data. One method commonly used in classifying images is the Neural Network method. In a Neural Network there are several activation functions that can be used in classifying, namely Tangent Hyperbolic and Rectified Linear Units. To find out which activation function is more effective in classifying, it is necessary to make a comparison between the two types of performance of this function. The dataset used to compare the performance of this function is a dataset of grape leaf images, which total 800 images and are grouped into 4 classes: 200 healthy images, 200 black-measles images, 200 black-rott images and 200 Isariopsis images. The results of this learning process show that TanH and ReLu can classify with the following Test and Score results: TanH with AUC 0.899, CA 0.735, F1 0.735, Precision 0.735, Recall 0.735 and MCC 0.647. ReLu with AUC 0.893, CA 0.713, F1 0.712, Precision 0.712, Recall 0.712 and MCC 0.617. From these results it can be concluded that in this study TanH performed better classification than ReLu in testing with this grape leaf image dataset.

Actions (login required)

View Item