公开(公告)号：US20180120916A1

公开(公告)日：2018-05-03

申请号：US15373466

申请日：2016-12-09

Applicant: Industrial Technology Research Institute

Inventor： Yung-Chieh Lin ,  Shih-Che Lin ,  Chao-Hong Chen ,  Liang-Chia Cheng

IPC: G06F1/32

CPC classification number: G06F1/324 ,  G06F1/3287

Abstract: A FPGA-based system power estimation apparatus and a method for estimating the power of a target intellectual property (IP) circuit are provided. The system power estimation apparatus includes a FPGA and a power analysis circuit. The FPGA is configured to accommodate the target IP circuit. The power analysis circuit is disposed into the FPGA. The power analysis circuit retrieves an internal operation-state signal of the target IP circuit. The power analysis circuit examines the internal operation-state signal to determine an operation state of the target IP circuit and uses a power model to convert the operation state of the target IP circuit into at least one power value.

公开(公告)号：US10324517B2

公开(公告)日：2019-06-18

申请号：US15373466

申请日：2016-12-09

Applicant: Industrial Technology Research Institute

Inventor： Yung-Chieh Lin ,  Shih-Che Lin ,  Chao-Hong Chen ,  Liang-Chia Cheng

IPC: G06F1/32 ,  G06F1/324 ,  G06F1/3287

Abstract: A FPGA-based system power estimation apparatus and a method for estimating the power of a target intellectual property (IP) circuit are provided. The system power estimation apparatus includes a FPGA and a power analysis circuit. The FPGA is configured to accommodate the target IP circuit. The power analysis circuit is disposed into the FPGA. The power analysis circuit retrieves an internal operation-state signal of the target IP circuit. The power analysis circuit examines the internal operation-state signal to determine an operation state of the target IP circuit and uses a power model to convert the operation state of the target IP circuit into at least one power value.

公开(公告)号：US11537893B2

公开(公告)日：2022-12-27

申请号：US16729480

申请日：2019-12-30

Applicant: Industrial Technology Research Institute

Inventor： Ming-Chun Hsyu ,  Chao-Hong Chen ,  Chien-Chih Huang

IPC: G06N3/08 ,  G06N3/04 ,  G06N20/00 ,  G06K9/62

Abstract: A method and an electronic device for selecting deep neural network hyperparameters are provided. In an embodiment of the method, a plurality of testing hyperparameter configurations are sampled from a plurality of hyperparameter ranges of a plurality of hyperparameters. A target neural network model is trained by using a training dataset and the plurality of testing hyperparameter configurations, and a plurality of accuracies corresponding to the plurality of testing hyperparameter configurations are obtained after training for preset epochs. A hyperparameter recommendation operation is performed to predict a plurality of final accuracies of the plurality of testing hyperparameter configurations. A recommended hyperparameter configuration corresponding to the final accuracy having a highest predicted value is selected as a hyperparameter setting for continuing training the target neural network model.

公开(公告)号：US20210174210A1

公开(公告)日：2021-06-10

申请号：US16729480

申请日：2019-12-30

Applicant: Industrial Technology Research Institute

Inventor： Ming-Chun Hsyu ,  Chao-Hong Chen ,  Chien-Chih Huang

IPC: G06N3/08 ,  G06N3/04 ,  G06N20/00 ,  G06K9/62

Abstract: A method and an electronic device for selecting deep neural network hyperparameters are provided. In an embodiment of the method, a plurality of testing hyperparameter configurations are sampled from a plurality of hyperparameter ranges of a plurality of hyperparameters. A target neural network model is trained by using a training dataset and the plurality of testing hyperparameter configurations, and a plurality of accuracies corresponding to the plurality of testing hyperparameter configurations are obtained after training for preset epochs. A hyperparameter recommendation operation is performed to predict a plurality of final accuracies of the plurality of testing hyperparameter configurations. A recommended hyperparameter configuration corresponding to the final accuracy having a highest predicted value is selected as a hyperparameter setting for continuing training the target neural network model.