METHOD AND DEVICE FOR SUPPORTING UNPROFESSIONAL WORKERS WHOSE TASK PROFICIENCIES ARE LOW TO READILY GENERATE TRAINING DATASETS BY USING SEMI-SUPERVISED LEARNING ALGORITHMS

摘要

Learning data that can be used to learn a main neural network designed to achieve a given purpose by non-professional workers included in a low-skilled group whose work proficiency is below a threshold, using a supporting neural network using a semi-supervised learning algorithm A method for supporting the creation of a set easily is disclosed. That is, (a) the computing device causes the support neural network to use the reference training data corresponding to at least one specific original data included in the original dataset used to generate the training dataset, the quasi-instruction generating preliminary learning data corresponding to the remaining original data except for the specific original data included in the original dataset by performing a learning algorithm; (b) the computing device, to each of the non-professional workers included in the low-skilled group, uses at least one of a preset first optimization format to an N-th optimization format, wherein N is an integer equal to or greater than 1; obtaining correction data corresponding to at least a portion of the preliminary learning data from the workers by providing at least a portion of the learning data; (c) supporting, by the computing device, to generate the training dataset with reference to the preliminary training data and the correction data.