Jing J. Liang, Hao Guo, Kunjie Yu, B. Qu, C. Yue, Kangjia Qiao
{"title":"An Improved Composite Differential Evolutionary Algorithm with Self-adaptive Mutation Strategy for Identifying Photovoltaic Model Parameters","authors":"Jing J. Liang, Hao Guo, Kunjie Yu, B. Qu, C. Yue, Kangjia Qiao","doi":"10.1109/acait53529.2021.9731329","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731329","url":null,"abstract":"With the rapid growth of solar energy demand, the optimization of the photovoltaic model becomes significant. The conversion efficiency of the photovoltaic model is mainly determined by its structural parameters, and the multi-modal property of parameter search space brings challenges to the existing evolutionary algorithms. Therefore, this paper proposes an improved composite differential evolutionary algorithm with a self-adaptive mutation strategy (CoDESA). In CoDESA, three complementary strategies are selected into the strategy pool, and each parent will produce three offspring according to their selection probabilities. Moreover, the selection probability of each strategy is dynamically adjusted using a self-adaptive mechanism, so that the algorithm can utilize the more suitable strategies at specific evolutionary stages. The proposed CoDESA is examined on the parameter identification of three photovoltaic models. It is compared with seven commonly used evolutionary algorithms, and more accurate parameters are identified.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125405374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mingyuan Wang, Jianjun Yuan, Sheng Bao, Liang Du, Shugen Ma
{"title":"Robots for pipeline inspection tasks—A survey of design philosophy and implementation technologies","authors":"Mingyuan Wang, Jianjun Yuan, Sheng Bao, Liang Du, Shugen Ma","doi":"10.1109/acait53529.2021.9731152","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731152","url":null,"abstract":"Pipeline system is one of the most economic and efficient methods for transporting fluids around the world. The document review shows that several robotic solutions have been developed to navigate inside or outside pipelines. But, most of these solutions are limited to specific scenarios. Based on the survey, this paper undergoes with the characteristics, application and challenges for development of pipe robots. A comprehensive categorization of pipe robots mainly focusing on modular design, bio-inspired and soft structure are addressed. The survey results show that existing out-pipe robotic solutions require more flexibility to adapt to such confined staggered environments and further research is necessary.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126306640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Research on Music Emotion Classification Based on CNN-LSTM Network","authors":"Yin Yu","doi":"10.1109/acait53529.2021.9731277","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731277","url":null,"abstract":"Music art contains rich emotional information. The research on the classification of music emotion is of great significance for massive music organization and retrieval. In view of this, this study extracts the feature parameters in music information based on support vector machine (SVM), convolutional neural network (CNN) and cyclic neural network (RNN). While analyzing the impact of different feature parameters on music emotion classification, this paper constructs a CNN-LSTM combined network classification model. The results show that compared with the traditional classification algorithms, the combined model constructed in this study has higher classification accuracy and can improve the performance of music emotion classification thoroughly.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121181838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Research on Face Recognition Technology Fusion Deep Learning Under Different Light Intensity Changes","authors":"Yanqing Yang, Xing Song","doi":"10.1109/acait53529.2021.9731292","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731292","url":null,"abstract":"Aiming at the problem of face recognition under different illumination intensities combined with deep learning algorithms, this research designed a new type of loss function, the I-center loss function. Use face image data set LFW with different light intensity to train and test LeNets++ deep learning network based on softmax, center, I-center loss function, and a variety of common image recognition networks. The calculation results show that although the LeNets++ deep learning network training requires much more data than other networks selected in the study, when the loss function is changed to I-center, the network has a significant improvement in the accuracy of face image recognition under different light intensities, reaching 99.65%. Therefore, experiments have proved that the use of an improved deep learning neural network based on the I-center loss function can improve the face recognition effect under different light intensities.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"83 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121664487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Intelligent Detection of Underwater Fish Speed Characteristics Based on Deep Learning","authors":"Xianghui Li, Xin Xia, Zhuhua Hu, Bingtao Han, Yaochi Zhao","doi":"10.1109/acait53529.2021.9731159","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731159","url":null,"abstract":"At present, the breeding area of Hainan province is 58,000 hectares, and the breeding industry is an important economic source of Hainan province. As an important breeding object in Hainan province, the daily activities and abnormal behaviors of the fish have a direct impact on the breeding yield and the breeding income. For mariculture fish, changes in behaviour are often reflected in the important behavioral feature of swimming speed. Fishes swim at different speeds when they are in different situation. Of course, the change of fish speed is not only related to their own behavior and health state, but also related to the water quality. When the water quality changes or the fish are subjected to some abnormal stimulation, fish swimming speed will change. Therefore, accurate and rapid acquisition of fish swimming speed can not only reflect the change of fish behavior intuitively, but also reflect the water quality to a certain extent, which is of great significance to the large breeding province. Based on this, in this paper, tracking algorithm combined YOLOv5 deep learning network and Kalman filter is used to conduct intelligent detection of the speed characteristics of underwater fish, and track and calculate the speed of a single fish, a number of fish and fish swarm respectively. The experimental results show that the tracking algorithm proposed in this paper can track the underwater fish and calculate the corresponding speed well.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"340 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122754549","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Liying Cheng, Xiaowei Wang, Dan Zhang, Longtao Jiang
{"title":"Face Recognition Algorithm Based on Broad Learning System","authors":"Liying Cheng, Xiaowei Wang, Dan Zhang, Longtao Jiang","doi":"10.1109/acait53529.2021.9731145","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731145","url":null,"abstract":"Face recognition is a well-known issue in the realm of image processing, which has made tremendous strides in recent years, owing to the rapid development of artificial intelligence technology, and has become one of the most prominent research areas in a variety of fields. However, when uncontrollable variables such as light, face occlusion, and expression change are present, the recognition accuracy suffers as a result of the change in facial features. Face recognition in a complex environment is challenging since the accuracy of the algorithm is insufficient. This paper proposes a k-means clustering face recognition method based on the Broad Learning System (BLS),and discusses the principle and performance of the algorithm. The experimental results demonstrate that the proposed strategy improves identification accuracy and is more resistant to noise interference without requiring any changes to the model structure.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114666205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Li Fang, Long Ye, Xinglong Ma, Ruiqi Wang, Wei Zhong, Qin Zhang
{"title":"Frame-Level Multiple Sound Sources Localization Based on Visual Understanding","authors":"Li Fang, Long Ye, Xinglong Ma, Ruiqi Wang, Wei Zhong, Qin Zhang","doi":"10.1109/acait53529.2021.9731148","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731148","url":null,"abstract":"Sound source localization is an important field of audio and visual research. In the dynamic performance stage, finding the positions of multiple sounding objects in real time can give the audience an immersive feeling. Due to the complexity of the performance scene, it is a challenge to perform audio-visual recognition and localization because of the audio overlapping and visual object masking. To address this problem, we propose a novel two-stream learning framework that disentangles different classes of audio-visual representations from complex scenes, then maps the audio area of each visual in multi-instance labels learning through adaptive multi-stream fusion, and localizes sounding instrument from coarse to fine. We have obtained the state-of-the-art results on the public dataset. Experiment results show that our method can effectively realize frame-level multiple sound sources location.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"325 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115295450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Research on Performance Test of Chinese Character APP Software Based on OL-ADE Algorithm","authors":"Chao Wang, Chenguang Zhao, Quanshun Fu","doi":"10.1109/acait53529.2021.9731282","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731282","url":null,"abstract":"Entering the information age, Chinese character APP assumes the task of cultural dissemination in a novel form as a carrier. And, software performance testing, a means to ensure software quality, is needed to maintain its performance. In view of this, the research first designs an opposition-based learning of adaptive evolution based on reverse learning strategy (OL-ADE), and then builds a Chinese character APP software performance test model on this basis and put it into application, and finally verify its effect through experiments. The results show that when the population size is 50 or 100, the OL-ADE algorithm has the least number of iterations under different data input ranges, and the number of iterations is within [40, 150]. The branch coverage is 100%, and the average number of iterations required by OL-ADE algorithm is the smallest. The above results show that OL-ADE algorithm has the best overall performance among the four algorithms, and can complete the software performance test of Chinese character APP.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115303785","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"An multi-task head pose estimation algorithm","authors":"Heng Song, Tianbao Geng, Maoli Xie","doi":"10.1109/acait53529.2021.9731346","DOIUrl":"https://doi.org/10.1109/acait53529.2021.9731346","url":null,"abstract":"Estimating head pose is a hot topic in facial behavior analysis and understanding. Most of the existing methods called two-stage method take head pose estimation and face detection as two separate tasks. In general, independent face boxes need to be proposed before head pose estimation. Such scheme is inefficient and has poor robustness. The existing estimation methods for head pose is lack of effective anti-noise design. In this paper, we propose a multi-task deep learning method, which integrate face detection and pose estimation together. Three kind of anti-interference strategy are proposed. Compared with the existing two-stage method, the proposed method can be performed with less consumption of resource. Benefited from the complementary characteristics of multi task joint learning, our proposed has higher accuracy. Experiments on several public datasets fully show that the attitude angle estimation error accuracy of our one stage algorithm reaches 1.96° (MAE). It is better than the existing state of the art method. The speed is twice as fast as that of the two-stage method.","PeriodicalId":173633,"journal":{"name":"2021 5th Asian Conference on Artificial Intelligence Technology (ACAIT)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123152150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}