APPARATUS FOR UNDERSTANDING COMMUNICATION MESSAGE BY USING RECURRENT NEURAL NETWORK, AND METHOD THEREFOR

WIPO Patent Application WO/2020/085567

A1

A method for understanding a communication message, according to one aspect of the technical idea of the present invention, can comprise the steps of: receiving a stream including a plurality of input characters included in a transmitted communication message; calculating a first prediction result which is the result predicted according to a first character of the stream; comparing the calculated first prediction result with a second character of the stream; calculating, on the basis of the comparison result with the first prediction result, a second prediction result diverging from a different point when the calculated first prediction result is different from the second character, and storing the first prediction result as a first stack; comparing the calculated second prediction result with the second character; and calculating, on the basis of the comparison result with the second prediction result, a third prediction result diverging from a different point when the calculated second prediction result is different from the second character, and storing the second prediction result as a second stack.

KIM JAE HOON (KR)
LEE KUN HEE (KR)
YI YUNG (KR)

PCT/KR2018/014463

April 30, 2020

November 22, 2018

Click for automatic bibliography generation  

UNIV AJOU IND ACADEMIC COOP FOUND (KR)
KOREA ADVANCED INST SCI & TECH (KR)

G06N3/08; H04L29/06

KR100891358B1	2009-04-01
KR20150036671A	2015-04-07
JP5447373B2	2014-03-19
JP2003076392A	2003-03-14

BANG, SUNG HYU K ET AL.: "Approach for Learning Intention Prediction Model based on Recurrent Neural Network", TRANSACTIONS OF INFORMATION SCIENTISTS AND ENGINEERS, vol. 45, no. 4, April 2018 (2018-04-01), pages 360 - 369, Retrieved from the Internet

CHUNG, Sungjoon (KR)

View/Download PDF      PDF Help

Previous Patent: FLEXIBLE LED FILM MODULE

Next Patent: ELECTRICALLY-DRIVEN, SELF-PROPELLED OIL SKIMMER