Text

METODE ENSEMBLE LEARNING TEKNIK WEIGHTED VOTING PADA ARSITEKTUR ALEXNET, VGG-16 DAN XCEPTION DALAM KLASIFIKASI PENYAKIT KANKER PAYUDARA

---

Penilaian

---

Automatic classification of breast cancer is done by utilizing ultrasound images (USG). Automatic early detection of breast cancer can use several CNN architectures such as AlexNet, VGG-16 and Xception. AlexNet is easy to implement, but less effective in capturing complex features. VGG-16 has a deep layer that can extract complex features, but has the risk of overfitting due to a too large number of parameters. The xception has ability to capture complex features like VGG-16 and produces fewer parameters because it uses depthwise separable convolution, but is vulnerable to outliers. Each architecture has advantages and disadvantages, to obtain better performance, ensemble learning can be used. This research method performs weighted voting by involving NASNetMobile learning to combine the weight results from AlexNet, VGG-16 and Xception. The application of NASNetMobile to the ensemble learning method is used to ensure that the weights obtained are optimal weights because NASNetMobile has the ability to adjust the model using the Neural Architecture Search approach. This method is applied to the classification of breast cancer in 3 classes, namely benign, malignant, and normal. The results of this method obtained the average accuracy, sensitivity, specificity, F1-Score and Cohen's Kappa as 96%, 96%, 98%, 96% and 0.94. The performance results above 90% show that the ensemble learning method of weighted voting technique with NASNetMobile learning is very good in classifying breast cancer. Classification results on the malignant class were excellent with performance above 96%, exceeding that of the normal and benign classes. Although the performance on the normal and benign classes was lower, both still reached above 90%. The results of the application of ensemble learning on average can improve the performance of AlexNet, VGG-16 and Xception by 5%.

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2507001056	T167328	T1673282025	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1673282025

Publisher

: Prodi Ilmu Matematika, Fakultas Matematika Dan Ilmu Pengetahuan Alam Universitas Sriwijaya., 2025

Collation

viii, 86 hlm.; ilus., tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

510.07

Content Type

-

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Matematika

Specific Detail Info

-

Statement of Responsibility

EM

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment