Processing

Please wait...

Settings

Settings

Goto Application

1. WO2020091259 - IMPROVEMENT OF PREDICTION PERFORMANCE USING ASYMMETRIC TANH ACTIVATION FUNCTION

Publication Number WO/2020/091259
Publication Date 07.05.2020
International Application No. PCT/KR2019/013316
International Filing Date 11.10.2019
IPC
G06N 3/08 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
G06N 3/04 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architecture, e.g. interconnection topology
Applicants
  • 에스케이텔레콤 주식회사 SK TELECOM CO., LTD. [KR]/[KR]
Inventors
  • 한용희 HAN, Yong Hee
Agents
  • 이철희 LEE, Chulhee
Priority Data
10-2018-012958729.10.2018KR
Publication Language Korean (KO)
Filing Language Korean (KO)
Designated States
Title
(EN) IMPROVEMENT OF PREDICTION PERFORMANCE USING ASYMMETRIC TANH ACTIVATION FUNCTION
(FR) AMÉLIORATION DES PERFORMANCES DE PRÉDICTION EN UTILISANT UNE FONCTION D'ACTIVATION EN TANH ASYMÉTRIQUE
(KO) 비대칭 TANH 활성 함수를 이용한 예측 성능의 개선
Abstract
(EN)
Provided is an asymmetric hyperbolic tanh function which can be used as an activation function irrespective of the structure of a neural network, according to one aspect of the present invention. The proposed activation function limits an output range thereof to between a maximum value and a minimum value of a variable to be predicted. The proposed activation function is suitable for a regression problem which requires the prediction of a wide range of real values on the basis of input data. Representative drawing: figure 3 Representative drawing: figure 3
(FR)
La présente invention concerne, selon un de ses aspects, une fonction de tangente hyperbolique asymétrique qui peut être utilisée comme fonction d'activation indépendamment de la structure d'un réseau neuronal. La fonction d'activation proposée limite une plage de sortie de celle-ci à celle comprise entre une valeur maximum et une valeur minimum d'une variable à prédire. La fonction d'activation proposée convient pour un problème de régression nécessitant la prédiction d'une plage étendue de valeurs réelles d'après des données d'entrée. Dessin représentatif: figure 3 Dessin représentatif: figure 3
(KO)
본 발명의 일 측면에 의하면, 뉴럴 네트워크의 구조에 상관 없이 활성 함수(activation function)로 사용가능한 비대칭의 하이퍼볼릭 탄젠트 함수(asymmetric tanh function)를 제공한다. 제안된 활성 함수는 그 출력 범위를 예측하고자 하는 변수의 최대값과 최소값 사이로 제한한다. 제안된 활성 함수는 입력 데이터에 따라 넓은 범위의 실수값을 예측해야 하는 회귀 문제에 적합하다. 대표도: 도 3
Latest bibliographic data on file with the International Bureau