US 11,941,513 B2
Device for ensembling data received from prediction devices and operating method thereof
Myung-Eun Lim, Daejeon (KR); Jae Hun Choi, Daejeon (KR); and Youngwoong Han, Daejeon (KR)
Assigned to ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE, Daejeon (KR)
Filed by ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE, Daejeon (KR)
Filed on Nov. 28, 2019, as Appl. No. 16/699,060.
Claims priority of application No. 10-2018-0156420 (KR), filed on Dec. 6, 2018; and application No. 10-2019-0124616 (KR), filed on Oct. 8, 2019.
Prior Publication US 2020/0184284 A1, Jun. 11, 2020
Int. Cl. G06N 3/08 (2023.01); G06F 18/24 (2023.01); G06F 18/25 (2023.01); G06N 3/045 (2023.01); G06N 3/063 (2023.01); G06N 20/20 (2019.01); G06V 10/82 (2022.01)
CPC G06N 3/063 (2013.01) [G06F 18/24 (2023.01); G06F 18/251 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06N 20/20 (2019.01); G06V 10/82 (2022.01)] 18 Claims
OG exemplary drawing
 
12. A device for ensembling data received from a plurality of prediction devices, the device comprising:
a data manager configured to:
provide time series data to a first prediction device and a second prediction device,
receive a first device prediction result corresponding to the time series data from the first prediction device, and
receive a second device prediction result corresponding to the time series data from the second prediction device; and
a predictor configured to: generate first item weights depending on first item values of the first and second device prediction results, second item weights depending on second item values of the first and second device prediction results, a first device weight corresponding to the first prediction device, and a second device weight corresponding to the second prediction device, based on the first and second device prediction results, and configured to generate an ensemble result of the first and second device prediction results, based on the first and second item weights and the first and second device weights,
wherein the predictor includes:
a preprocessor configured to preprocess the first and second item values of the first and second device prediction results;
an item weight calculator configured to calculate the first item weights, based on the preprocessed first item values, and calculate the second item weights, based on the preprocessed second item values; and
a device weight calculator configured to calculate the first and second device weights, based on the preprocessed first and second device prediction results, and
wherein the item weight calculator includes a neural network that includes:
a first input layer configured to receive the first item values;
a first embedding layer configured to output a first intermediate result in consideration of a relationship between the first item values; and
a first attention layer configured to analyze the first intermediate result and calculate the first item weights using an attention mechanism.