-
公开(公告)号：US20240232632A1
公开(公告)日：2024-07-11
申请号：US18343303
申请日：2023-06-28
Applicant: SAMSUNG ELECTRONICS CO., LTD.
Inventor： Gunhee KIM , Hyuntae CHO
Abstract: A processor-implemented method may include generating respective first neural network differential data by differentiating a respective output of each layer of a first neural network with respect to input data provided to the first neural network that estimates output data from the input data, by a forward propagation process of the first neural network, generating, using a second neural network, an output differential value of the output data with respect to the input data using the respective first neural network differential data, and training the first neural network and the second neural network based on ground truth data of the output data and ground truth data of the output differential value.
-
公开(公告)号：US20230017432A1
公开(公告)日：2023-01-19
申请号：US17950342
申请日：2022-09-22
Applicant: Samsung Electronics Co., Ltd.
Inventor： Sangwon HA , Gunhee KIM , Donghyun LEE
Abstract: A processor-implemented neural network method includes: determining a respective probability density function (PDF) of normalizing a statistical distribution of parameter values, for each channel of each of a plurality of feature maps of a pre-trained neural network; determining, for each channel, a corresponding first quantization range for performing quantization of corresponding parameter values, based on a quantization error and a quantization noise of the respective determined PDF; determining, for each channel, a corresponding second quantization range, based on a signal-to-quantization noise ratio (SQNR) of the respective determined PDF; correcting, for each channel, the corresponding first quantization range based on the corresponding second quantization range; and generating a quantized neural network, based on the corrected first quantization range corresponding for each channel.
-
公开(公告)号：US20230169333A1
公开(公告)日：2023-06-01
申请号：US17862881
申请日：2022-07-12
Applicant: SAMSUNG ELECTRONICS CO., LTD.
Inventor： CHANGIN CHOI , Gunhee KIM , YONGDEOK KIM , MYEONG WOO KIM , SEUNGWON LEE , NARANKHUU TUVSHINJARGAL
IPC: G06N3/08
CPC classification number: G06N3/08
Abstract: Disclosed are a training method and apparatus for distributed training of a neural network, the training apparatus including processors configured to perform distributed training, wherein each of the processors is further configured to perform a forward direction operation for layers of the neural network, determine a loss of the neural network based on the forward direction operation, determine a local gradient for each layer of the neural network by performing a backward direction operation for the layers of the neural network based on the loss, determine whether to perform gradient clipping for a local gradient determined for a previous layer, in response to determining a local gradient for a current layer through the backward direction operation, determine an aggregated gradient based on the backward direction operation and the gradient clipping performed by each of the processors, and update parameters of the neural network based on the aggregated gradient.
-
公开(公告)号：US20220012588A1
公开(公告)日：2022-01-13
申请号：US17348678
申请日：2021-06-15
Applicant: SAMSUNG ELECTRONICS CO., LTD. , SNU R&DB FOUNDATION
Inventor： Seonmin RHEE , Chris Dongjoo KIM , Gunhee KIM , Jinseo JEONG , Seungju HAN
Abstract: A reservoir management method includes: in response to receiving input data to which label information is mapped, determining whether to add the input data to a reservoir based on a sampling probability; in response to determining to add the input data to the reservoir when the reservoir is filled, selecting candidate data to be removed from among sets of sample data included in the reservoir based on a target label distribution and a current label distribution of the reservoir, and removing the selected candidate data from the reservoir; and training a neural network model using sample data of the reservoir from which the selected candidate data is removed.
-
公开(公告)号：US20230140239A1
公开(公告)日：2023-05-04
申请号：US17868361
申请日：2022-07-19
Applicant: Samsung Electronics Co., Ltd.
Inventor： Myeong Woo KIM , Yongdeok KIM , Narankhuu TUVSHINJARGAL , Gunhee KIM , Seungwon LEE , Changin CHOI
IPC: G06F16/906 , G06F40/20 , G06K9/62
Abstract: A processor-implemented method with data loading includes: dividing a training data set into a plurality of subsets based on sizes of a plurality of data files included in the training data set; loading, from each of the plurality of subsets, a portion of data files in the subset to a plurality of processors based on a proportion of a number of data files of the plurality of subsets in the subset and a batch size of distributed training; and reallocating, based on sizes of data files loaded to processors in a same group among the plurality of processors, the loaded data files to the processors in the same group.
-
Patent Agency Ranking
公开(公告)号：US20220092384A1
公开(公告)日：2022-03-24
申请号：US17192048
申请日：2021-03-04
Applicant: SAMSUNG ELECTRONICS CO., LTD.
Inventor： Gunhee KIM , Seungwon LEE
IPC: G06N3/04
Abstract: A method of quantizing parameters of a neural network includes acquiring a parameter of a floating-point format used in a process of inferring by the neural network, quantizing, based on statistics of a weight included in the parameter, the weight into a fixed-point format, determining, based on statistics of an activation of one or more layers configuring the neural network included in the parameter, a dynamic range of the activation, and quantizing, based on statistics of input data of the neural network, the input data into a fixed-point format.
-
公开(公告)号：US20220027668A1
公开(公告)日：2022-01-27
申请号：US17340134
申请日：2021-06-07
Inventor： Seon Min RHEE , Jaekyeom KIM , Gunhee KIM , Minjung KIM , Dongyeon WOO , Seungju HAN
Abstract: A neural network includes a drop layer configured to drops feature values. A method of computation using the neural network includes extracting feature data from input data using a first portion of a neural network, generating compressed representation data of the extracted feature data by dropping a feature value from the extracted feature data at a drop layer of the neural network based on a drop probability corresponding to the feature value, and indicating an inference result from the compressed representation data using a second portion of the neural network.
-
公开(公告)号：US20210201117A1
公开(公告)日：2021-07-01
申请号：US16909095
申请日：2020-06-23
Applicant: Samsung Electronics Co., Ltd.
Inventor： Sangwon HA , Gunhee KIM , Donghyun LEE
Abstract: A processor-implemented neural network method includes: determining a respective probability density function (PDF) of normalizing a statistical distribution of parameter values, for each channel of each of a plurality of feature maps of a pre-trained neural network; determining, for each channel, a corresponding first quantization range for performing quantization of corresponding parameter values, based on a quantization error and a quantization noise of the respective determined PDF; determining, for each channel, a corresponding second quantization range, based on a signal-to-quantization noise ratio (SQNR) of the respective determined PDF; correcting, for each channel, the corresponding first quantization range based on the corresponding second quantization range; and generating a quantized neural network, based on the corrected first quantization range corresponding for each channel.
-
公开(公告)号：US20210136467A1
公开(公告)日：2021-05-06
申请号：US17046494
申请日：2019-03-27
Inventor： Sunghyun KIM , Yongdeok KIM , Gunhee KIM , Joonil NA , Jinyoung SUNG , Youngjae YU , Sangho LEE
IPC: H04N21/8549 , G06N3/08 , G06F16/583 , G06F16/55 , H04N21/8547 , G06F16/58
Abstract: Provided are an electronic device and an operation method thereof. The electronic device includes a memory that stores one or more instructions, and a processor that executes the one or more instructions stored in the memory, wherein the processor is configured to execute the one or more instructions to: divide original image data into a plurality of image sequences; determine a predetermined number of image sequences among the plurality of image sequences as an input image group, select one of the image sequences included in the input image group and add the selected image sequence to the highlight image group based on one or more image sequences pre-classified as a highlight image group, by using a trained model trained using an artificial intelligence algorithm; and generate summary image data extracted from the original image data, by using the image sequence included in the highlight image group.
-
公开(公告)号：US20230306262A1
公开(公告)日：2023-09-28
申请号：US18187030
申请日：2023-03-21
Applicant: SAMSUNG ELECTRONICS CO., LTD.
Inventor： Gunhee KIM , Seungwon LEE
IPC: G06N3/08 , G06N3/0464
CPC classification number: G06N3/08 , G06N3/0464
Abstract: A processor-implemented method is provided. The method includes, for each layer of a plurality of layers of a neural network for an input data provided to the neural network, obtain activation data of a corresponding layer of the plurality of layers, resulting from an inference operation of the corresponding layer; generate differential data of the activation data of the corresponding layer with respect to input data; and generate differential data of output data of the neural network with respect to the input data, based on the generated differential data of each layer.
-
-
-
-
-
-
-
-
-
Search Results
11
records
(took 0.02 seconds)
