With the rapid development of smart grid, the substation secondary cable condition monitoring data is growing exponentially and gradually constitutes the secondary circuit condition monitoring big data. The traditiona...
With the rapid development of smart grid, the substation secondary cable condition monitoring data is growing exponentially and gradually constitutes the secondary circuit condition monitoring big data. The traditional computing architecture can no longer meet the computing performance demand. Combining Spark big data processing technology and AliCloud E-MapReduce cloud computing platform, we propose a parallel patternrecognition method for substation secondary cable condition monitoring big data, aiming to improve the ability of the secondary cable online monitoring system to quickly batch analyze the alarm monitoring data that suddenly increase in a short period of time. Spark-KNN, a parallelized K-nearest neighbor classification algorithm based on Spark, is designed to realize parallel patternrecognition of massive secondary cable monitoring data. The experimental results show that the average performance of Spark-KNN is 3.17 times higher than that of Hadoop MapReduce implementation, which is more suitable for performing real-time processing tasks of secondary cable monitoring big data.
Image classification has been a trendy research topic in the field of patternrecognition and computer vision, which extracts different features of images and predict the category of images. Thanks to the development ...
详细信息
Image classification has been a trendy research topic in the field of patternrecognition and computer vision, which extracts different features of images and predict the category of images. Thanks to the development of deep learning, powerful convolutional neural networks can be used in the field of image recognition. However, existing deep learning-based image recognition research mostly follows the framework of supervised learning, and model learning relies on a large number of accurate labels. Providing a large amount of label data will undoubtedly require laborious human effort and expensive costs. Therefore, image recognition based on unsupervised learning (without using any image category labels to achieve classification) has become a spotlight for research. In this paper, explorations on the image classification by self-supervised framework SimCLR on image classification successfully clusters a large number of images into an optimum amount categories. Qualitative results have show SimCLR is particularly effective in recognizing the colors of images; both qualitative and quantitative results shows SimCLR is great at identifying simple contours. However, when the colors are similar, and contour lines are complex, SimCLR does not obtain satisficing results. The accuracy on classifying Mnist dataset is 32%.
This study addresses the crucial task of architectural decorative image patternrecognition in the context of iconography, with an emphasis on efficient information mining. The proposed research work presents a novel ...
详细信息
ISBN:
(数字)9798350396157
ISBN:
(纸本)9798350396164
This study addresses the crucial task of architectural decorative image patternrecognition in the context of iconography, with an emphasis on efficient information mining. The proposed research work presents a novel algorithm that combines edge detection, ResNet50, and the triplet attention mechanism for image style analysis. The proposed methodology is organized into comprehensive sections, including a literature review that explores various studies in image patternrecognition. The proposed algorithm focuses on overcoming the challenges faced by existing deep learning methods in capturing aesthetic style features. It introduces a multifaceted self-supervised task and incorporates edge detection using the Canny method, followed by the application of ResNet50 with a triplet attention mechanism. The paper then focuses on the optimization of the network structure for the targeted task of architectural decorative image recognition, introducing a feature embedding encoder (FEE) to effectively handle multi-level structures. In the experimental phase, the proposed model is tested against traditional CNN and FCM models, demonstrating superior performance with recognition accuracy consistently ranging from 97% to 99%. The comparative analysis highlights the effectiveness of the proposed algorithm in achieving high accuracy, positioning it as a promising solution for image recognition tasks.
In order to identify arrhythmia more simply and efficiently, one-dimensional convolutional neural network is used to extract electrocardiogram signal features and a neural network model based on cardiac beat is establ...
详细信息
To solve the problem of low accuracy of gait recognition in complex scenes, a novel skeleton-based gait recognition algorithm, GCGait, is proposed. Taking human posture as the input of gait feature, the interference c...
To solve the problem of low accuracy of gait recognition in complex scenes, a novel skeleton-based gait recognition algorithm, GCGait, is proposed. Taking human posture as the input of gait feature, the interference caused by wearing changes and other factors is reduced. To extract sufficient input features, multi-branch input is used in the early stage of the model. By introducing the multi-attention mechanism, the network can learn the semantic information of the non-directly connected joints, excavate the most discriminative features from complex videos, and further improve the recognition performance. In order to reduce the influence of cross view, the fusion loss function is used in the experiment. Experimental results show that the average recognition rate of the proposed algorithm on the CASIA-B dataset is improved by 5.2%, and the average recognition accuracy on the OU-MVLP dataset is increased by 66.3%, which proves the effectiveness of the proposed method.
Based on multi-label load identification, power load feature analysis and intelligent identification method based on Symmetrized Dot pattern (SDP) information fusion. This method improves the decomposition efficiency ...
详细信息
ISBN:
(数字)9798350375138
ISBN:
(纸本)9798350375145
Based on multi-label load identification, power load feature analysis and intelligent identification method based on Symmetrized Dot pattern (SDP) information fusion. This method improves the decomposition efficiency and solves the problem of large reconstruction error; proposes the method of SDP fusion feature analysis, extracts the SDP image feature of each modal information of the load, improves the completeness of the information; proposes the SDP image recognition method based on YOLOv5 and builds on load intelligent recognition model. Through experimental research, the load identification accuracy of this method reached 98%, which ensured the level of non-invasive load monitoring.
This study proposes a deep learning-based genome sequence function prediction model and systematically evaluates its performance. The study begins with an overview of the importance of genome sequence function predict...
详细信息
ISBN:
(数字)9798331519032
ISBN:
(纸本)9798331519049
This study proposes a deep learning-based genome sequence function prediction model and systematically evaluates its performance. The study begins with an overview of the importance of genome sequence function prediction and its applications in biomedicine, followed by a detailed description of the main current research methods and advances. We use Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Transformer model for capturing local features, sequence dependencies and long-range dependencies, respectively. Through the combined application of multiple feature extraction methods, the model demonstrates excellent prediction performance on different genomic datasets. The experimental results show that the deep learning model can effectively improve the accuracy of genomic sequence function prediction, especially in complex feature extraction and patternrecognition. Convolutional neural networks have obvious advantages in local feature extraction, recurrent neural networks excel in dealing with the time dependence of sequence data, and the Transformer model demonstrates its superiority in dealing with long-distance dependencies and parallel computation.
Active deception jamming is one of the common means to jam radar signals. How to effectively recognize active deception jamming is a challenge of modern radar technology. To address the accuracy and real-time of radar...
详细信息
Machine Learning (ML) has been widely applied to medical science for decades. As common knowledge, the progress of many diseases is often chronic and dynamic. Longitudinal data, or time-series data, has better descrip...
详细信息
The Inter-Class Word Similarities in combination with Intra-Class Variations make it a difficult task for an OCR or any other machine learning system to recognize the handwritten characters and words with high accurac...
详细信息
暂无评论