The patent badge is an abbreviated version of the USPTO patent document. The patent badge does contain a link to the full patent document.

The patent badge is an abbreviated version of the USPTO patent document. The patent badge covers the following: Patent number, Date patent was issued, Date patent was filed, Title of the patent, Applicant, Inventor, Assignee, Attorney firm, Primary examiner, Assistant examiner, CPCs, and Abstract. The patent badge does contain a link to the full patent document (in Adobe Acrobat format, aka pdf). To download or print any patent click here.

Date of Patent:
Apr. 20, 2021

Filed:

Jun. 20, 2019
Applicant:

Sogang University Research Foundation, Seoul, KR;

Inventors:

Myoung-Wan Koo, Seoul, KR;

Byoungjae Kim, Seoul, KR;

Jung Yun Seo, Seoul, KR;

Attorney:
Primary Examiner:
Int. Cl.
CPC ...
G10L 15/16 (2006.01); G10L 15/197 (2013.01); G10L 15/22 (2006.01); G06N 3/04 (2006.01); G06N 3/08 (2006.01); G06F 16/33 (2019.01); G10L 15/30 (2013.01);
U.S. Cl.
CPC ...
G10L 15/197 (2013.01); G06F 16/3347 (2019.01); G06N 3/0445 (2013.01); G06N 3/08 (2013.01); G10L 15/16 (2013.01); G10L 15/22 (2013.01); G10L 15/30 (2013.01); G10L 2015/223 (2013.01);
Abstract

Provided is a voice conversation method using a bi-directional LSTM (Bi-LSTM) memory network. The voice conversation method includes steps of: performing bi-directional LSTM processing on history sentences to generate first sentence embedding and second sentence embedding; performing bi-directional LSTM processing on a question sentence to generate question sentence embedding; inputting the first sentence embedding and the question sentence embedding to an input memory of a memory network and inputting the second sentence embedding into an output memory of the memory network; generating a weight matrix representing relevance between the first sentence embedding of the input memory and the question sentence embedding, performing weighted sum of the weight matrix and the second sentence embedding of the output memory to generate output information, adding the output information and the question sentence embedding to obtain prediction information, and outputting the prediction information; and selecting an answer sentence by using the prediction information.


Find Patent Forward Citations

Loading…