Pengukuran akurasi image classification menggunakan metode convolutional neural network pada tulisan angka mandarin

Erlina, Erlina (2022) Pengukuran akurasi image classification menggunakan metode convolutional neural network pada tulisan angka mandarin. Bachelor thesis, Universitas Pelita Harapan.

Preview	Text (Title) Title.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (234kB) | Preview
Preview	Text (Abstract) Abstract.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (291kB) | Preview
Preview	Text (ToC) ToC.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (645kB) | Preview
Preview	Text (Chapter1) Chapter 1.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (689kB) | Preview
	Text (Chapter2) Chapter 2.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (5MB)
	Text (Chapter3) Chapter 3.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (3MB)
	Text (Chapter4) Chapter 4.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter5) Chapter 5.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (272kB)
Preview	Text (Bibliography) Bibliography.pdf Available under License Creative Commons Attribution Non-commercial Share Alike. Download (366kB) | Preview
	Text (Appendices) Appendices.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Share Alike. Download (939kB)

Abstract

Thanks to China’s rapid economic growth, has an impact on the increasing interest and popularity of Mandarin in the world. This has caused Mandarin to succeed in winning the position of being one of the languages that have an important role after English. Each character in Chinese consists of strokes that have their own meaning so that if one stroke is missing, the meaning of the character will also change. The length of the strokes also has an important role in determining the meaning of the character. If one of the strokes has a longer stroke, then the meaning will also change. Therefore, categorizing Chinese characters is more difficult when compared to other languages that use Latin characters. In this case, AI can be applied to help categorize Chinese characters with a lower error rate than humans. In this research, a Machine Learning model will be built using the Convolutional Neural Network method to perform the Image Classification process. The results of using the Convolutional Neural Network algorithm on the model built to provide an accuracy rate of 99%. / Berkat melesatnya pertumbuhan ekonomi negara Tiongkok berdampak pada meningkatnya minat dan popularitas Bahasa Mandarin di dunia. Hal ini menyebabkan Bahasa Mandarin berhasil menyabet posisi menjadi salah satu bahasa yang memiliki peran penting setelah Bahasa Inggris. Tiap karakter pada Bahasa Mandarin terdiri atas goresan-goresan yang memiliki maknanya sendiri sehingga apabila ada 1 goresan yang kurang maka makna dari karakter tersebut pun juga akan berubah. Bukan hanya jumlah goresan yang memiliki peran penting dalam karakter Bahasa Mandarin tetapi juga panjang dari goresan tersebut oleh karena itu, mengkategorikan karakter Bahasa Mandarin cukup sulit jika dibandingkan dengan bahasa lainnya yang menggunakan huruf latin. Dalam permasalahan ini, AI dapat diaplikasikan untuk membantu mengkategorikan karakter Bahasa Mandarin dengan tingkat kesalahan yang lebih rendah dibandingkan dengan manusia. Dalam penelitian ini akan dibangun model Machine Learning menggunakan metode Convolutional Neural Network untuk melakukan proses Image Classification. Hasil dari aplikasi algoritma Convolutional Neural Network pada model yang dibangun memberikan tingkat akurasi sebesar 99%.

Actions (login required)

View Item