DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Yoo, Hoi-Jun | - |
dc.contributor.advisor | 유회준 | - |
dc.contributor.author | Kim, Changhyeon | - |
dc.date.accessioned | 2021-05-12T19:45:34Z | - |
dc.date.available | 2021-05-12T19:45:34Z | - |
dc.date.issued | 2020 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=924538&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/284450 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 전기및전자공학부, 2020.8,[iv, 47 p. :] | - |
dc.description.abstract | Recently, deep neural network (DNN) has been actively researched from simple recognition tasks to precise control for robot or autonomous systems, which are treated as a task that only human can do. Unlike recognition tasks, the real-time operation is essential in action control, and it is too slow to use remote learning on a server communicating through a network. New learning techniques, such as reinforcement learning (RL), are needed to determine and select the correct robot behavior locally. In this paper, we propose a low power deep reinforcement learning (DRL) SoC, supporting CNN and learning-optimized RNNs. The adaptive reusability of weights and inputs, and data encoding/decoding techniques reduces power consumption and peak memory bandwidth of DRL processing by 31% and 41%, respectively. The 65nm 16mm2 chip achieves a peak 2.16TFLOPS/W at 0.73V and 204 GFLOPS at 1.1V with 16b data. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | 딥러닝▼a딥러닝ASIC▼a모바일 딥러닝▼a인공신경망▼a강화학습 | - |
dc.subject | deep Learning▼adeep learning ASIC▼amobile deep learning▼adeep neural network▼areinforcement learning | - |
dc.title | (An) energy-efficient deep reinforcement learning SoC for mobile platform | - |
dc.title.alternative | 모바일 플랫폼용 저전력 딥러닝 기반 강화학습 가속 SoC 설계 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 김창현 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.