Error-correcting output codes based on Hadamard matrices and deep neural network
- 주제어 (키워드) error-correcting output codes(ECOC) , Hadamard matrices , convolutional neural network(CNN)
- 발행기관 서강대학교 일반대학원
- 지도교수 김종락
- 발행년도 2022
- 학위수여년월 2022. 2
- 학위명 석사
- 학과 및 전공 일반대학원 수학과
- 실제 URI http://www.dcollection.net/handler/sogang/000000066511
- UCI I804:11029-000000066511
- 본문언어 영어
- 저작권 서강대학교 논문은 저작권 보호를 받습니다.
초록 (요약문)
The error-correcting output codes(ECOC) is one of the ensemble methods for the multi-class classification problem. Hadamard matrices are used because they have good properties for the ECOC coding matrix. In this thesis, we apply Hadamard matrices of various orders to Hadamard ECOC to determine which orders of Hadamard matrices give a good performance. For our experiment, a convolutional neural network(CNN) is chosen as a base learner, which is largely used in image classification problems.
more