Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238062
Junwen Tong, Jiabao Zhao, Jiaoyang Jin, Weisong Qiao, Haitong Li
PointNet extracts global shape features from the unordered point sets directly, respecting the permutation invariance of the input points; however, it fails to capture the fine-grained local shape features. In this paper, we extend PointNet to a multi-input and single-output structure by additionally feeding the scale-invariant heat kernel signature into PointNet to capture the fine-grained local shape features. To diversify the training data, we resample the points of each model randomly and generate a set of sub-samples, based on which PointNet calculates their classification scores. Then we adopt a plurality voting strategy to fuse the sub-sample level feature vectors to a model level descriptor, according to their classification scores. The experimental results demonstrate our proposed method outperforms the state-of-the-art retrieval methods on two 3D model benchmarks.
{"title":"An Improved Deep Multiple-input and Single-output PointNet for 3D Model Retrieval","authors":"Junwen Tong, Jiabao Zhao, Jiaoyang Jin, Weisong Qiao, Haitong Li","doi":"10.1109/ICNSC48988.2020.9238062","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238062","url":null,"abstract":"PointNet extracts global shape features from the unordered point sets directly, respecting the permutation invariance of the input points; however, it fails to capture the fine-grained local shape features. In this paper, we extend PointNet to a multi-input and single-output structure by additionally feeding the scale-invariant heat kernel signature into PointNet to capture the fine-grained local shape features. To diversify the training data, we resample the points of each model randomly and generate a set of sub-samples, based on which PointNet calculates their classification scores. Then we adopt a plurality voting strategy to fuse the sub-sample level feature vectors to a model level descriptor, according to their classification scores. The experimental results demonstrate our proposed method outperforms the state-of-the-art retrieval methods on two 3D model benchmarks.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"35 6","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114031563","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238069
Yuwei Liang, Huaxiong Li, Bing Huang, Zhuohuai Guan, Pei Yang
This paper aims to develop a novel cost-sensitive face recognition framework, which can gain the desirable recognition results with the least total cost. By combining two recently rising techniques: deep convolutional neural networks (CNNs) and sequential three-way decision (3WD) method, our framework can automatically label new samples and incorporates the delayed decision into decision-making process. We first explore the semi-supervised face recognition method in the case of the scarcity of labeled training data. By learning the class estimation and the deep convolution feature extraction of the unlabeled data jointly, the CNN trained by both labeled training data and unlabeled data is generated. Then, rather than getting a lower recognition error rate, we focus on seeking the minimum cost of misclassification at each decision step. For this purpose, we introduce the method of sequential 3WD in our cost-sensitive face recognition framework, which take each iteration of semi-supervised learning as a decision-making step. When there are insufficient labeled samples, a delayed decision will be adopted to reduce the decision cost. Finally, the test cost is also considered in the decision-making process, and the sum of misclassification cost and test cost is taken as the total cost. Using the total cost as the objective function, optimizing the performance indicators, training to get the classifier with the smallest total cost. In short, the model strives to get an optimal decision step, so that the reliable identification result can be obtained with only a small number of labeled data. The work value of this paper is to prove the effectiveness of our method in two face datasets.
{"title":"Semi-Supervised Incremental Three-Way Decision Using Convolutional Neural Network","authors":"Yuwei Liang, Huaxiong Li, Bing Huang, Zhuohuai Guan, Pei Yang","doi":"10.1109/ICNSC48988.2020.9238069","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238069","url":null,"abstract":"This paper aims to develop a novel cost-sensitive face recognition framework, which can gain the desirable recognition results with the least total cost. By combining two recently rising techniques: deep convolutional neural networks (CNNs) and sequential three-way decision (3WD) method, our framework can automatically label new samples and incorporates the delayed decision into decision-making process. We first explore the semi-supervised face recognition method in the case of the scarcity of labeled training data. By learning the class estimation and the deep convolution feature extraction of the unlabeled data jointly, the CNN trained by both labeled training data and unlabeled data is generated. Then, rather than getting a lower recognition error rate, we focus on seeking the minimum cost of misclassification at each decision step. For this purpose, we introduce the method of sequential 3WD in our cost-sensitive face recognition framework, which take each iteration of semi-supervised learning as a decision-making step. When there are insufficient labeled samples, a delayed decision will be adopted to reduce the decision cost. Finally, the test cost is also considered in the decision-making process, and the sum of misclassification cost and test cost is taken as the total cost. Using the total cost as the objective function, optimizing the performance indicators, training to get the classifier with the smallest total cost. In short, the model strives to get an optimal decision step, so that the reliable identification result can be obtained with only a small number of labeled data. The work value of this paper is to prove the effectiveness of our method in two face datasets.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"517 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133565766","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238067
Shanshan Yuan
The small-sample problems in scientific research can be settled by the information diffusion in one-dimensional Euclidean space. In this paper, the diffusion in two-dimensional Euclidean space has been developed, according to the author's previous study. It has also made obvious achievements in the applications of the two-dimensional technology to the two practical cases in healthcare.
{"title":"Small-Sample Information Diffusion and Applications","authors":"Shanshan Yuan","doi":"10.1109/ICNSC48988.2020.9238067","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238067","url":null,"abstract":"The small-sample problems in scientific research can be settled by the information diffusion in one-dimensional Euclidean space. In this paper, the diffusion in two-dimensional Euclidean space has been developed, according to the author's previous study. It has also made obvious achievements in the applications of the two-dimensional technology to the two practical cases in healthcare.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115162647","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238056
C. Zhang, Shuanghua Yang
An efficient time synchronisation could increase the performance of most wireless sensor network (WSN) applications significantly. The energy conservation of sensor nodes using the theory of periodic activation and sleep relies on schedule tables with accurate timestamps which are maintained by frequent time synchronisations. In this paper, a time synchronisation scheme with a new Medium Access Control (MAC) protocol, named Schedule Awareness MAC (SA-MAC), is proposed for energy conservation purposes. The research focuses on MAC and application layer protocols. In SA-MAC, every sensor node obtains its two-hop neighbour active time slots information by applying dimensionality reduction to generate concentrated schedule tables. The simulation results show the lifetime of the network is extended significantly with an acceptable time error rate.
{"title":"A Lightweight Time Synchronisation for Wireless Sensor Networks","authors":"C. Zhang, Shuanghua Yang","doi":"10.1109/ICNSC48988.2020.9238056","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238056","url":null,"abstract":"An efficient time synchronisation could increase the performance of most wireless sensor network (WSN) applications significantly. The energy conservation of sensor nodes using the theory of periodic activation and sleep relies on schedule tables with accurate timestamps which are maintained by frequent time synchronisations. In this paper, a time synchronisation scheme with a new Medium Access Control (MAC) protocol, named Schedule Awareness MAC (SA-MAC), is proposed for energy conservation purposes. The research focuses on MAC and application layer protocols. In SA-MAC, every sensor node obtains its two-hop neighbour active time slots information by applying dimensionality reduction to generate concentrated schedule tables. The simulation results show the lifetime of the network is extended significantly with an acceptable time error rate.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133189631","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238058
Z. Liu, Xianzhong Zhou, Wenting Hu
With the continuous improvement of China's economy and life, the hotel industry is booming simultaneously. As a major management business for hotels, purchasing management of room consumables is crucial. In reality, hotels usually purchase multiple types of room consumables through supply middlemen, and there may be combined effects between these raw material products of consumables during the procurement process. This type of combined effects can enable hotels to increase efficiency and reduce costs in purchasing activities, and thereby influence their choice of manufacturer. Therefore, by studying the combined effect of purchasing room consumables and selecting the optimal manufacturer, the purpose of reducing hotel procurement costs and increasing revenue is achieved. This paper proposes a manufacturer selection algorithm for hotel room consumables based on combined effects. The raw material products can be combined into several groups with similar attributes by analyzing the combined degree of raw material products. Finally, the optimal manufacturer can be determined which meets the requirements based on the attributes of the combined groups.
{"title":"Selection of raw material manufacturers for hotel room consumables based on combined effects","authors":"Z. Liu, Xianzhong Zhou, Wenting Hu","doi":"10.1109/ICNSC48988.2020.9238058","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238058","url":null,"abstract":"With the continuous improvement of China's economy and life, the hotel industry is booming simultaneously. As a major management business for hotels, purchasing management of room consumables is crucial. In reality, hotels usually purchase multiple types of room consumables through supply middlemen, and there may be combined effects between these raw material products of consumables during the procurement process. This type of combined effects can enable hotels to increase efficiency and reduce costs in purchasing activities, and thereby influence their choice of manufacturer. Therefore, by studying the combined effect of purchasing room consumables and selecting the optimal manufacturer, the purpose of reducing hotel procurement costs and increasing revenue is achieved. This paper proposes a manufacturer selection algorithm for hotel room consumables based on combined effects. The raw material products can be combined into several groups with similar attributes by analyzing the combined degree of raw material products. Finally, the optimal manufacturer can be determined which meets the requirements based on the attributes of the combined groups.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121848752","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238104
Xu Han, Rongbai Zhang, Xingzi Liu, Frank Jiang
With the increase of Internet usage, the identification and recovery from cyber-attacks become the major concerns for cyber industries. Therefore, the harm caused by network attacks has caused widespread concern. Distributed Denial of Service (DDoS) attack is a very common destructive cyber attack. This is a network attack that destroys the network and can cause multiple computers to be attacked at the same time, failing to perform services properly. Therefore, based on the understanding of blockchain structure and DDoS characteristics, a blockchain-based DDoS detection model framework is proposed to form a blockchain-based collaborative detection system. We use the blockchain consortium chain structure to treat all participants as part of the private chain in the system. Each participating organization has its own channel, and other organizations cannot access its information, thus fully protecting the privacy of each participant. Our experimental results show that smart contracts can detect DDoS data and generate anomalous chains on each node. The time required to generate an exception chain and information sharing is very short, which indicates that the system can protect the privacy of user data. While sharing data in time, good results can be obtained as a collaborative detection system.
{"title":"Biologically Inspired Smart Contract: A Blockchain-Based DDoS Detection System","authors":"Xu Han, Rongbai Zhang, Xingzi Liu, Frank Jiang","doi":"10.1109/ICNSC48988.2020.9238104","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238104","url":null,"abstract":"With the increase of Internet usage, the identification and recovery from cyber-attacks become the major concerns for cyber industries. Therefore, the harm caused by network attacks has caused widespread concern. Distributed Denial of Service (DDoS) attack is a very common destructive cyber attack. This is a network attack that destroys the network and can cause multiple computers to be attacked at the same time, failing to perform services properly. Therefore, based on the understanding of blockchain structure and DDoS characteristics, a blockchain-based DDoS detection model framework is proposed to form a blockchain-based collaborative detection system. We use the blockchain consortium chain structure to treat all participants as part of the private chain in the system. Each participating organization has its own channel, and other organizations cannot access its information, thus fully protecting the privacy of each participant. Our experimental results show that smart contracts can detect DDoS data and generate anomalous chains on each node. The time required to generate an exception chain and information sharing is very short, which indicates that the system can protect the privacy of user data. While sharing data in time, good results can be obtained as a collaborative detection system.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129281759","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238108
Furui Wang, Xuemin Chen, G. Song
In this paper, to detect bolt looseness of a subsea flange, we develop a new percussion method using speech recognition technology and least square support vector machine. Especially, to extract features from percussion-induced sound signals, we employ the mel frequency cepstral coefficient (MFCC). Finally, an experiment is conducted to verify the effectiveness of the proposed method. Compared to current detection methods for bolt loosening, the proposed method can avoid constant contact between sensors and structures, which significantly improves practicability and provides guidance for structural health monitoring based on the cyber-physics systems.
{"title":"Percussion-based Detection of Bolt Looseness Using Speech Recognition Technology and Least Square Support Vector Machine","authors":"Furui Wang, Xuemin Chen, G. Song","doi":"10.1109/ICNSC48988.2020.9238108","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238108","url":null,"abstract":"In this paper, to detect bolt looseness of a subsea flange, we develop a new percussion method using speech recognition technology and least square support vector machine. Especially, to extract features from percussion-induced sound signals, we employ the mel frequency cepstral coefficient (MFCC). Finally, an experiment is conducted to verify the effectiveness of the proposed method. Compared to current detection methods for bolt loosening, the proposed method can avoid constant contact between sensors and structures, which significantly improves practicability and provides guidance for structural health monitoring based on the cyber-physics systems.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124638434","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238105
Xiaodong Huang, Lei Peng, Cheng Lu, J. Bi, Haitao Yuan
In light curves (LCs), the brightness of stars is associated with time, and so is its image. The traditional data processing methods cannot effectively handle real-time and large-volume data of various LCs. To address this issue, this work develops a deep neural network named Dropout-based Recurrent Neural Networks (DRNN). It extracts complicated features of all images captured by Mini Ground-based Wide-Angle Camera array (Mini-GWAC) for point source extraction and cross-certification through Long Short-Term Memory units. DRNN can also produce warnings for abnormal values of light change curves. Furthermore, this work optimizes the training model by combining a dropout method with an adaptive moment estimation algorithm to iteratively update the network weight of the RNN based on the LCs data. Extensive experiments with a Mini-GWAC dataset demonstrate that DRNN outperforms several typical methods in terms of prediction performance of star brightness in large-scale astronomical LCs.
{"title":"A Deep Learning Approach to Large-Scale Light Curve Prediction and Real-Time Anomaly Detection with Grubbs Criterion","authors":"Xiaodong Huang, Lei Peng, Cheng Lu, J. Bi, Haitao Yuan","doi":"10.1109/ICNSC48988.2020.9238105","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238105","url":null,"abstract":"In light curves (LCs), the brightness of stars is associated with time, and so is its image. The traditional data processing methods cannot effectively handle real-time and large-volume data of various LCs. To address this issue, this work develops a deep neural network named Dropout-based Recurrent Neural Networks (DRNN). It extracts complicated features of all images captured by Mini Ground-based Wide-Angle Camera array (Mini-GWAC) for point source extraction and cross-certification through Long Short-Term Memory units. DRNN can also produce warnings for abnormal values of light change curves. Furthermore, this work optimizes the training model by combining a dropout method with an adaptive moment estimation algorithm to iteratively update the network weight of the RNN based on the LCs data. Extensive experiments with a Mini-GWAC dataset demonstrate that DRNN outperforms several typical methods in terms of prediction performance of star brightness in large-scale astronomical LCs.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121273546","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238125
Guanyu Cai, Yuqin Wang, Lianghua He, Mengchu Zhou
Transfer learning, especially unsupervised domain adaptation, is a crucial technology for sample-efficient learning. Recently, deep adversarial domain adaptation methods perform remarkably well in various tasks, which introduce a domain classifier to promote domain-invariant representation. However, previous methods either constrain the representative ability with an identical feature extractor for both domains or ignore the relationship between domains with separate extractors. In this paper, we propose a novel adversarial domain adaptation method named Adversarial Transform Network (ATN) to both enhance the representative ability and transfer general information between domains. Residual connections are used to share features in the bottom layers, which deliver transferrable features to boost generalization performance. Moreover, a regularizer is proposed to alleviate a vanishing gradient problem, thus stabilizing the optimization procedure. Extensive experiments are conducted to show that the proposed ATN is comparable with the methods of the state-of-the-art and effectively deals with the vanishing gradient problem.
{"title":"Adversarial Transform Networks for Unsupervised Transfer Learning","authors":"Guanyu Cai, Yuqin Wang, Lianghua He, Mengchu Zhou","doi":"10.1109/ICNSC48988.2020.9238125","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238125","url":null,"abstract":"Transfer learning, especially unsupervised domain adaptation, is a crucial technology for sample-efficient learning. Recently, deep adversarial domain adaptation methods perform remarkably well in various tasks, which introduce a domain classifier to promote domain-invariant representation. However, previous methods either constrain the representative ability with an identical feature extractor for both domains or ignore the relationship between domains with separate extractors. In this paper, we propose a novel adversarial domain adaptation method named Adversarial Transform Network (ATN) to both enhance the representative ability and transfer general information between domains. Residual connections are used to share features in the bottom layers, which deliver transferrable features to boost generalization performance. Moreover, a regularizer is proposed to alleviate a vanishing gradient problem, thus stabilizing the optimization procedure. Extensive experiments are conducted to show that the proposed ATN is comparable with the methods of the state-of-the-art and effectively deals with the vanishing gradient problem.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126945764","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}
Pub Date : 2020-10-30DOI: 10.1109/ICNSC48988.2020.9238081
Huiqing Zhang, Shuo Li, Donghao Li
Sonar technology plays an important role in the development of marine resources and military strategy. Due to the bad underwater acoustic channel, the sonar image collected by sonar technology equipment is affected by various kinds of distortions easily. To obtain high-quality sonar image, we devise a novel dual-path deep neural network (DPDNN) to measure the quality of sonar image. In these two paths, we use the batch normalization layer to reduce the training time and take the skip operation to speed up the feature extraction. Based on the above two operations, we extract the micro-scopic and macro-scopic structure of sonar image, respectively. Finally, the global average pooling layer and the fully connection layer are used to connect the above two paths. Experiments show that our DPDNN has a significant improvement in prediction performance and efficiency, respectively.
{"title":"A Dual-Path Deep Neural Network for Sonar Image Quality Evaluation","authors":"Huiqing Zhang, Shuo Li, Donghao Li","doi":"10.1109/ICNSC48988.2020.9238081","DOIUrl":"https://doi.org/10.1109/ICNSC48988.2020.9238081","url":null,"abstract":"Sonar technology plays an important role in the development of marine resources and military strategy. Due to the bad underwater acoustic channel, the sonar image collected by sonar technology equipment is affected by various kinds of distortions easily. To obtain high-quality sonar image, we devise a novel dual-path deep neural network (DPDNN) to measure the quality of sonar image. In these two paths, we use the batch normalization layer to reduce the training time and take the skip operation to speed up the feature extraction. Based on the above two operations, we extract the micro-scopic and macro-scopic structure of sonar image, respectively. Finally, the global average pooling layer and the fully connection layer are used to connect the above two paths. Experiments show that our DPDNN has a significant improvement in prediction performance and efficiency, respectively.","PeriodicalId":412290,"journal":{"name":"2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125324685","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：