DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Yoo, Hoi-Jun | - |
dc.contributor.advisor | 유회준 | - |
dc.contributor.author | Kim, Youngwoo | - |
dc.date.accessioned | 2021-05-13T19:39:18Z | - |
dc.date.available | 2021-05-13T19:39:18Z | - |
dc.date.issued | 2020 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=925219&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/285055 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2020.8,[iii, 26 p. :] | - |
dc.description.abstract | A low-power, highly secure, always-on face recognition (FR) processor is required for security applications. In this paper, a branch net–based early stopping FR (BESF) processor is proposed to prevent adversarial attacks for high security and consume low power for always-on operation. It shows a recognition accuracy of 83.10% under the fast gradient signed method (FGSM), and 71.97% under the projected gradient descent (PGD) attack. The clock-gating of the BESF processor reduces the average power consumption by 30.85%. The unified pointwise and depthwise convolution processing element adopts layer-fusion to reduce the external memory access by 88.0%. Furthermore, noise injection layers are inserted between every bottleneck layer to further reduce the FGSM and PGD attack success rate by 9.29% and 20.0%, respectively. Implemented with a 65 nm CMOS process with a 3.0 mm $\times$ 3.0 mm area, the processor consumes 0.22–0.89 mW power at 1 fps and shows 95.5% FR accuracy in the Labeled Faces in the Wild dataset. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Adversarial attack▼aConvolutional neural network▼aExternal memory access▼aFace recognition▼aNoise injection | - |
dc.subject | 적대적 공격▼a합성곱 신경망▼a외부 메모리 접근▼a얼굴 인식▼a노이즈 삽입 | - |
dc.title | (A) low-power always-on CNN face recognition processor with adversarial attack prevention | - |
dc.title.alternative | 적대적 공격 방어 가능한 저전력 CNN 얼굴 인식 프로세서 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 김영우 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.