DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Chung, Sae-Young | - |
dc.contributor.advisor | 정세영 | - |
dc.contributor.author | Lee, Honghee | - |
dc.date.accessioned | 2022-04-27T19:31:37Z | - |
dc.date.available | 2022-04-27T19:31:37Z | - |
dc.date.issued | 2021 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=948970&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/296055 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2021.2,[iii, 16 p. :] | - |
dc.description.abstract | For machine learning systems, anomaly detection which detects samples that are vastly different from the dataset is essential for achieving robust training and reliable results. Although it has been actively studied for the last several years, performing accurate detection on high-dimensional data remains a challenge. Conventional reconstruction-based methods have shown good performance, but they rely on the heuristic that reconstruction error of the anomalous data is larger than that of the normal data. Thus, these methods are not trained by optimizing a detection based objective function and show suboptimal performances. To tackle this problem, we propose an autoencoder-based method which is capable of regularizing the latent space for better detection by scoring anomalies through a distance metric. The proposed method leverages the additional distance loss function which makes the data close to the center along with the original reconstruction loss function. The effectiveness of our method is evaluated on several high-dimensional network intrusion detection datasets. Our method outperforms existing reconstruction-based methods and is also robust to hyperparameter selection. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | anomaly detection▼areconstruction▼aautoencoder network▼adistance metric▼adistance loss function | - |
dc.subject | 이상 현상 탐지▼a복원▼a오토인코더 네트워크▼a거리 메트릭▼a거리 오차 함수 | - |
dc.title | Reconstruction based anomaly detection using distance metric | - |
dc.title.alternative | 거리 메트릭을 이용한 복원 기반 이상 현상 탐지 연구 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 이홍희 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.