DEEP LEARNING-BASED BITCOIN BLOCK DATA PREDICTION SYSTEM TAKING INTO ACCOUNT TIME SERIES DISTRIBUTION CHARACTERISTICS

WIPO Patent Application WO/2022/080583

A1

Disclosed are a deep learning-based Bitcoin block data prediction system and method that take into account time series distribution characteristics. The deep learning-based Bitcoin block data prediction system that takes into account time series distribution characteristics according to the present invention comprises: a data collection module for collecting a plurality of class data including block data, social media data, and price data; a pre-processing module which performs pre-processing for unifying the data formats of the plurality of collected class data as time series data, and clusters the time series data into time series data sets according to the distribution characteristics of the respective time series data for the plurality of pre-processed class data; a learning module which is trained on the plurality of clustered time series data sets through a deep learning-based model, and generates a plurality of prediction models according to the plurality of time series data sets; and a prediction module which evaluates the plurality of learned prediction models, and which, when new data is input, selects a prediction model with which to perform prediction from among the plurality of prediction models.

KIM MYUNG-SUP (KR)
BAEK EUIJUN (KR)
PARK JEETAE (KR)
KIM BOSEON (KR)

PCT/KR2020/017919

April 21, 2022

December 09, 2020

Click for automatic bibliography generation  

UNIV KOREA RES & BUSINESS FOUNDATION SEJONG CAMPUS (KR)

G06Q10/06; G06F40/30; G06Q10/04; G06Q20/06

KR20200115708A	2020-10-08
KR101862000B1	2018-05-29
KR20190013038A	2019-02-11
KR20200036219A	2020-04-07
KR20200005206A	2020-01-15

YANG, Sungbo (KR)

View/Download PDF      PDF Help

Previous Patent: TARGET-ORIENTED REINFORCEMENT LEARNING METHOD, AND DEVICE FOR PERFORMING SAME

Next Patent: MULTISTAGE LOAD-ADJUSTABLE MOVING CART