The language communication of social robots has a great significance meaning for improving the robot's intelligence and humanity.This research explores the impact of the dialogue type and voice parameter settings of social robots in the education industry on their perceived personality traits and acceptance, and constructs the principles and models of voice applications for robots that are suitable for the education industry.The study included the influence of the educational robot's dialogue type and speech parameters, gender , fundamental frequency, and speech rate. Through the control of robot's speech and dialogue content, to measure the personality of the robot during the interaction between the subjects and the robots. The experiment invited 32 subjects to conduct VHRI robot experiments, which was found that there is a significant interactive effects among the dialogue types and voice parameters of social robots used in the education industry. In addition, the gender of sound and fundamental frequency can improve the personal characteristics of robot extroversion (E), agreeableness (A),conscientiousness (C), openness (O), and anti-neurotic (N). Speech rates can improve the robot's agreeableness(A), conscientiousness (C), anti-neurotic (N), openness (O) and other characteristics. This study confirmed that the robot's dialogue types and voice parameter setting can be used to improve the robot's perceptual personality traits to improve its humanity. On this basis , the results will help social robots play their professional role in the application of education better and get better integrate into human society ,at the same time, it provides references for the voice parameters design of social robotics in different industries in the future.
One-class classification for model generation using only one sample category plays an important role in particular object recognition, missing category treatment and prediction of geographic event distributions. However, traditional one-class classification algorithm is difficult to achieve high accuracy in remote sensing because of existing larger amounts of spatial bands in remote sensing image as well as uncertainty or inconsistency information in relatively few training samples. To solve these problems, this study proposed a sparse and stratified sampling method for one-class classification of remote sensing images (SSSOCC). Spatial features were first divided into multiple groups which have different sample subspaces through sparse description and stratified sampling. Each group trained a one-class support vector machine as a sub-classifier, and then made joint decisions according to multiple sub-classifiers. Experiments show that our method's classification accuracy and stability are better than those of traditional one-class support vector machine. Thus, this study could meet the demands for one category land use or cover information extraction from remote sensing images. It could also solve problems such as larger amounts of spatial bands, uncertainty or inconsistency information in relatively few training samples.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.