US 11,868,199 B2
System for detecting anomaly and method for setting threshold thereof
Ji Hoon Kim, Seoul (KR); Sung Soo Park, Incheon (KR); Chang Woo Chun, Suwon-si (KR); Song Eun Lee, Seongnam-si (KR); Hyun Soo Cho, Seoul (KR); Jun Yeob Kim, Seoul (KR); and Sang Goo Lee, Seoul (KR)
Assigned to HYUNDAI MOTOR COMPANY, Seoul (KR); KIA CORPORATION, Seoul (KR); and SEOUL NATIONAL UNIVERSITY R&DS FOUNDATION, Seoul (KR)
Filed by HYUNDAI MOTOR COMPANY, Seoul (KR); Kia Corporation, Seoul (KR); and Seoul National University R&DB Foundation, Seoul (KR)
Filed on Mar. 23, 2022, as Appl. No. 17/702,234.
Claims priority of application No. 10-2021-0119157 (KR), filed on Sep. 7, 2021.
Prior Publication US 2023/0076888 A1, Mar. 9, 2023
Int. Cl. G06F 11/07 (2006.01); G06F 17/18 (2006.01)
CPC G06F 11/076 (2013.01) [G06F 17/18 (2013.01)] 14 Claims
OG exemplary drawing
 
1. A method for setting a threshold of a system for detecting an anomaly, the method comprising:
setting, by a processor, a minimum true positive rate (TPR) and a differentiation slope based on a user input;
creating, by the processor, a receiver operating characteristic (ROC) curve using anomaly data and normal data;
estimating, by the processor, a true negative rate (TNR) from the ROC curve based on the minimum TPR and the differentiation slope; and
estimating and setting, by the processor, the threshold based on the estimated TNR;
selecting, by the processor, one of at least one speech recognition engine based on the estimated threshold in response to receiving a speech signal from the outside using a microphone; and
performing, by the processor, speech recognition on the speech signal using the selected speech recognition engine,
wherein the threshold is automatically reset when a processing region of the at least one speech recognition engine is changed or a new speech recognition engine is introduced.