TransDIR: Deformable imaging registration network based on transformer to improve the feature extraction ability

2021 ◽  
Author(s):  
Tiejun Yang ◽  
Xinhao Bai ◽  
Xiaojuan Cui ◽  
Yuehong Gong ◽  
Lei Li
Keyword(s):  
Feature Extraction ◽  
Extraction Ability

scholarly journals Application of Deep Learning in Fault Diagnosis of Rotating Machinery

Processes ◽  
2021 ◽  
Vol 9 (6) ◽  
pp. 919
Author(s):  
Wanlu Jiang ◽  
Chenyang Wang ◽  
Jiayun Zou ◽  
Shuqing Zhang
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Fault Diagnosis ◽  
Rotating Machinery ◽  
Generative Adversarial Networks ◽  
Extraction Ability ◽  
One Dimensional ◽  
Adversarial Networks ◽  
Diagnosis Model ◽  
The One

The field of mechanical fault diagnosis has entered the era of “big data”. However, existing diagnostic algorithms, relying on artificial feature extraction and expert knowledge are of poor extraction ability and lack self-adaptability in the mass data. In the fault diagnosis of rotating machinery, due to the accidental occurrence of equipment faults, the proportion of fault samples is small, the samples are imbalanced, and available data are scarce, which leads to the low accuracy rate of the intelligent diagnosis model trained to identify the equipment state. To solve the above problems, an end-to-end diagnosis model is first proposed, which is an intelligent fault diagnosis method based on one-dimensional convolutional neural network (1D-CNN). That is to say, the original vibration signal is directly input into the model for identification. After that, through combining the convolutional neural network with the generative adversarial networks, a data expansion method based on the one-dimensional deep convolutional generative adversarial networks (1D-DCGAN) is constructed to generate small sample size fault samples and construct the balanced data set. Meanwhile, in order to solve the problem that the network is difficult to optimize, gradient penalty and Wasserstein distance are introduced. Through the test of bearing database and hydraulic pump, it shows that the one-dimensional convolution operation has strong feature extraction ability for vibration signals. The proposed method is very accurate for fault diagnosis of the two kinds of equipment, and high-quality expansion of the original data can be achieved.

scholarly journals A Lightweight Convolutional Neural Network Based on Visual Attention for SAR Image Target Classification

Sensors ◽  
2018 ◽  
Vol 18 (9) ◽  
pp. 3039 ◽  
Author(s):  
Jiaqi Shao ◽  
Changwen Qu ◽  
Jianwei Li ◽  
Shujuan Peng
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Visual Attention ◽  
Convolutional Neural Network ◽  
Recognition Accuracy ◽  
Target Recognition ◽  
Sar Image ◽  
Extraction Ability ◽  
Data Imbalance ◽  
Recognition Efficiency

With the continuous development of the convolutional neural network (CNN) concept and other deep learning technologies, target recognition in Synthetic Aperture Radar (SAR) images has entered a new stage. At present, shallow CNNs with simple structure are mostly applied in SAR image target recognition, even though their feature extraction ability is limited to a large extent. What’s more, research on improving SAR image target recognition efficiency and imbalanced data processing is relatively scarce. Thus, a lightweight CNN model for target recognition in SAR image is designed in this paper. First, based on visual attention mechanism, the channel attention by-pass and spatial attention by-pass are introduced to the network to enhance the feature extraction ability. Then, the depthwise separable convolution is used to replace the standard convolution to reduce the computation cost and heighten the recognition efficiency. Finally, a new weighted distance measure loss function is introduced to weaken the adverse effect of data imbalance on the recognition accuracy of minority class. A series of recognition experiments based on two open data sets of MSTAR and OpenSARShip are implemented. Experimental results show that compared with four advanced networks recently proposed, our network can greatly diminish the model size and iteration time while guaranteeing the recognition accuracy, and it can effectively alleviate the adverse effects of data imbalance on recognition results.

scholarly journals Improved U-Net3+ with Stage Residual for Brain Tumor Segmentation

2021 ◽  
Author(s):  
Chuanbo Qin ◽  
Yujie Wu ◽  
Wenbin Liao ◽  
Junying Zeng ◽  
Shufen Liang ◽  
...  
Keyword(s):  
Feature Extraction ◽  
Brain Tumor ◽  
Feature Fusion ◽  
Three Dimensional ◽  
Batch Size ◽  
Tumor Segmentation ◽  
Extraction Ability ◽  
Brain Tumor Segmentation ◽  
Residual Structure ◽  
3D Network

Abstract Background For the coding part of U-Net3+, the brain tumor feature extraction ability is insufficient, leading to insufficient feature fusion when sampling on the network and reducing the segmentation accuracy. Methods In this study, we propose an improved U-Net3+ segmentation network based on stage residual. In the encoder part, the encoder based on the stage residual structure is used to reduce the degradation problem caused by the increase in network depth and enhance the feature extraction ability of the encoder, which is convenient for full feature fusion when sampling on the network. Besides, we used a filter response normalization (FRN) layer instead of a batch normalization layer to eliminate batch size impact on the network. Based on the improved U-Net3+ two-dimensional (2D) model with stage residual, IResUnet3+ three-dimensional (3D) model is constructed. We explore appropriate methods to deal with 3D data, which achieve accurate segmentation of the 3D network. Results The experimental results showed that: the sensitivity of WT, TC, and ET increased by 1.34%, 4.6%, and 8.44%, respectively. And the Dice coefficients of ET and WT were further increased by 3.43% and 1.03%, respectively. To facilitate further research, source code can be found at: https://github.com/YuOnlyLookOne/IResUnet3Plus. Conclusion In the segmentation task of brain tumor brats2018 dataset, compared with the classical networks u-net, v-net, resunet and u-net3 +, the proposed network has smaller parameters and significantly improved accuracy.

scholarly journals Integration of Ensemble GoogLeNet and Modified Deep Residual Networks for Short-Term Load Forecasting

Electronics ◽  
2021 ◽  
Vol 10 (20) ◽  
pp. 2455
Author(s):  
Aijia Ding ◽  
Tingzhang Liu ◽  
Xue Zou
Keyword(s):  
Feature Extraction ◽  
Load Forecasting ◽  
Generalization Capability ◽  
Extraction Ability ◽  
Short Term ◽  
Proposed Model ◽  
Short Term Load Forecasting ◽  
Probabilistic Load ◽  
Electrical Demand ◽  
Public Datasets

Due to the strong volatility of the electrical load and the defect of a time-consuming problem, in addition to overfitting existing in published forecasting methods, short-term electrical demand is difficult to forecast accurately and robustly. Given the excellent capability of weight sharing and feature extraction for convolution, a novel hybrid method based on ensemble GoogLeNet and modified deep residual networks for short-term load forecasting (STLF) is proposed to address these issues. Specifically, an ensemble GoogLeNet with dense block structure is used to strengthen feature extraction ability and generalization capability. Meanwhile, a group normalization technique is used to normalize outputs of the previous layer. Furthermore, a modified deep residual network is introduced to alleviate a vanishing gradient problem in order to improve the forecasting results. The proposed model is also adopted to conduct probabilistic load forecasting with Monte Carlo Dropout. Two acknowledged public datasets are used to evaluate the performance of the proposed methodology. Multiple experiments and comparisons with existing state-of-the-art models show that this method achieves accurate prediction results, strong generalization capability, and satisfactory coverages for different prediction intervals, along with reducing operation times.

scholarly journals An Improved YOLOv2 for Vehicle Detection

Sensors ◽  
2018 ◽  
Vol 18 (12) ◽  
pp. 4272 ◽  
Author(s):  
Jun Sang ◽  
Zhongyuan Wu ◽  
Pei Guo ◽  
Haibo Hu ◽  
Hong Xiang ◽  
...  
Keyword(s):  
Feature Extraction ◽  
Vehicle Detection ◽  
Training Dataset ◽  
Validation Dataset ◽  
Detection Accuracy ◽  
Extraction Ability ◽  
Detection Model ◽  
Vehicle Type ◽  
Proposed Model ◽  
Bounding Boxes

Vehicle detection is one of the important applications of object detection in intelligent transportation systems. It aims to extract specific vehicle-type information from pictures or videos containing vehicles. To solve the problems of existing vehicle detection, such as the lack of vehicle-type recognition, low detection accuracy, and slow speed, a new vehicle detection model YOLOv2_Vehicle based on YOLOv2 is proposed in this paper. The k-means++ clustering algorithm was used to cluster the vehicle bounding boxes on the training dataset, and six anchor boxes with different sizes were selected. Considering that the different scales of the vehicles may influence the vehicle detection model, normalization was applied to improve the loss calculation method for length and width of bounding boxes. To improve the feature extraction ability of the network, the multi-layer feature fusion strategy was adopted, and the repeated convolution layers in high layers were removed. The experimental results on the Beijing Institute of Technology (BIT)-Vehicle validation dataset demonstrated that the mean Average Precision (mAP) could reach 94.78%. The proposed model also showed excellent generalization ability on the CompCars test dataset, where the “vehicle face” is quite different from the training dataset. With the comparison experiments, it was proven that the proposed method is effective for vehicle detection. In addition, with network visualization, the proposed model showed excellent feature extraction ability.

scholarly journals CSNN: An Augmented Spiking based Framework with Perceptron-Inception

2018 ◽  
Author(s):  
Qi Xu ◽  
Yu Qi ◽  
Hang Yu ◽  
Jiangrong Shen ◽  
Huajin Tang ◽  
...  
Keyword(s):  
Neural Networks ◽  
Feature Extraction ◽  
Feature Learning ◽  
Classification Performance ◽  
Learning Ability ◽  
Extraction Ability ◽  
Feature Maps ◽  
Learning Capabilities ◽  
Biological Realism ◽  
High Level

Spiking Neural Networks (SNNs) represent and transmit information in spikes, which is considered more biologically realistic and computationally powerful than the traditional Artificial Neural Networks. The spiking neurons encode useful temporal information and possess highly anti-noise property. The feature extraction ability of typical SNNs is limited by shallow structures. This paper focuses on improving the feature extraction ability of SNNs in virtue of powerful feature extraction ability of Convolutional Neural Networks (CNNs). CNNs can extract abstract features resorting to the structure of the convolutional feature maps. We propose a CNN-SNN (CSNN) model to combine feature learning ability of CNNs with cognition ability of SNNs.  The CSNN model learns the encoded spatial temporal representations of images in an event-driven way. We evaluate the CSNN model on the handwritten digits images dataset MNIST and its variational databases. In the presented experimental results, the proposed CSNN model is evaluated regarding learning capabilities, encoding mechanisms, robustness to noisy stimuli and its classification performance. The results show that CSNN behaves well compared to other cognitive models with significantly fewer neurons and training samples. Our work brings more biological realism into modern image classification models, with the hope that these models can inform how the brain performs this high-level vision task.

scholarly journals A Hierarchical Sparse Discriminant Autoencoder for Bearing Fault Diagnosis

2022 ◽  
Vol 12 (2) ◽  
pp. 818
Author(s):  
Mengjie Zeng ◽  
Shunming Li ◽  
Ranran Li ◽  
Jiantao Lu ◽  
Kun Xu ◽  
...  
Keyword(s):  
Feature Extraction ◽  
Fault Diagnosis ◽  
Network Performance ◽  
Data Sets ◽  
Extraction Ability ◽  
Data Set ◽  
Penalty Term ◽  
Bearing Fault Diagnosis ◽  
Label Information

Although some traditional autoencoders and their extensions have been widely used in the research of intelligent fault diagnosis of rotating parts, their feature extraction capabilities are limited without label information. In response to this problem, this research proposes a hierarchical sparse discriminant autoencoder (HSDAE) method for fault diagnosis of rotating components, which is a new semi-supervised autoencoder structure. By considering the sparsity of autoencoders, a hierarchical sparsity strategy was proposed to improve the stacked sparsity autoencoders, and the particle swarm optimization algorithm was used to obtain the optimal sparsity parameters to improve network performance. In order to enhance the classification of the autoencoder, a class aggregation and class separability strategy was used, which is an additional discriminative distance that was added as a penalty term in the loss function to enhance the feature extraction ability of the network. Finally, the reliability of the proposed method was verified on the bearing data set of Case Western Reserve University and the bearing data set of the laboratory test platform. The results of comparison with other methods show that the HSDAE method can enhance the feature extraction ability of the network and has reliability and stability for different data sets.

An Improved Quantitative Image Analyser

1977 ◽  
Vol 35 ◽  
pp. 226-227
Author(s):  
J.P. Fallon ◽  
P.J. Gregory ◽  
C.J. Taylor
Keyword(s):  
Feature Extraction ◽  
Quantitative Analysis ◽  
Frequency Content ◽  
General Sense ◽  
Physical Measurements ◽  
Quantitative Image ◽  
Image Analyser ◽  
Automatic Methods ◽  
Quantitative Results

Quantitative image analysis systems have been used for several years in research and quality control applications in various fields including metallurgy and medicine. The technique has been applied as an extension of subjective microscopy to problems requiring quantitative results and which are amenable to automatic methods of interpretation.Feature extraction. In the most general sense, a feature can be defined as a portion of the image which differs in some consistent way from the background. A feature may be characterized by the density difference between itself and the background, by an edge gradient, or by the spatial frequency content (texture) within its boundaries. The task of feature extraction includes recognition of features and encoding of the associated information for quantitative analysis.Quantitative Analysis. Quantitative analysis is the determination of one or more physical measurements of each feature. These measurements may be straightforward ones such as area, length, or perimeter, or more complex stereological measurements such as convex perimeter or Feret's diameter.

Sign in / Sign up

Export Citation Format

Share Document