Traditional Convolutional Neural Network Models Research Articles

Crop Disease Identification by Fusing Multiscale Convolution and Vision Transformer.

With the development of smart agriculture, deep learning is playing an increasingly important role in crop disease recognition. The existing crop disease recognition models are mainly based on convolutional neural networks (CNN). Although traditional CNN models have excellent performance in modeling local relationships, it is difficult to extract global features. This study combines the advantages of CNN in extracting local disease information and vision transformer in obtaining global receptive fields to design a hybrid model called MSCVT. The model incorporates the multiscale self-attention module, which combines multiscale convolution and self-attention mechanisms and enables the fusion of local and global features at both the shallow and deep levels of the model. In addition, the model uses the inverted residual block to replace normal convolution to maintain a low number of parameters. To verify the validity and adaptability of MSCVT in the crop disease dataset, experiments were conducted in the PlantVillage dataset and the Apple Leaf Pathology dataset, and obtained results with recognition accuracies of 99.86% and 97.50%, respectively. In comparison with other CNN models, the proposed model achieved advanced performance in both cases. The experimental results show that MSCVT can obtain high recognition accuracy in crop disease recognition and shows excellent adaptability in multidisease recognition and small-scale disease recognition.

Bearing Fault Diagnosis based on Residual Network Attention Mechanism

Aiming at the problem that the detection of bearing fault diagnosis is rarely applied in the research of image classification, a new method based on residual network and attention mechanism is proposed to identify bearing fault diagnosis. One-dimensional vibration signals are transformed into two-dimensional time-frequency images by continuous wavelet transform (CWT), which are input into the model for classification. In order to solve the problem that the traditional convolutional neural network model ignores the low diagnostic accuracy of channel attention and spatial attention due to the loss of important features, the attention mechanism CBAM module is added to make up for the loss of channel features and spatial features in the traditional model. At the same time, the residual network Resnet combined with the attention mechanism can better capture the global information of the time frequency graph, and make up for the defects of the residual network module. The experimental results show that the model has high diagnostic accuracy in rolling bearing fault diagnosis, which proves that the proposed method is effective and feasible.

Reconstructing Ocean Subsurface Temperature and Salinity from Sea Surface Information Based on Dual Path Convolutional Neural Networks

Satellite remote sensing can provide observation information of the sea surface, and using the sea surface information to reconstruct the subsurface temperature (ST) and subsurface salinity (SS) information has significant application values. This study proposes an intelligent algorithm based on Dual Path Convolutional Neural Networks (DP-CNNs) to reconstruct the ST and SS. The DP-CNN can integrate known information including sea surface temperature (SST), sea surface salinity (SSS), and sea surface height (SSH) to reconstruct the ST and SS. The reconstruction model based on DP-CNN can solve the problem of detail information loss in traditional CNN (Convolutional Neural Network) models. This study performs experiments for the South China Sea under different seasons using reanalysis data. The experimental results show that the DP-CNN models have higher reconstruction accuracy than the CNN models, and this proves that DP-CNNs effectively mitigate the loss of detailed information in the CNN models. Compared with the ground truth data, the ST/SS reconstruction results of the DP-CNN model exhibited a high coefficient of determination (0.93/0.86) and a low root mean square error (around 0.31 °C/0.05 PSU). Therefore, the DP-CNN models can be used as an effective approach to reconstruct ST and SS using sea surface information.

Movie Recommendation System Based on Traditional Recommendation Algorithm and CNN Model

As streaming services have expanded in recent years, an excellent recommendation algorithm, as one of the core technologies in movie service, brings huge benefits. Although the current application research of recommendation algorithms is mature, most systems or products usually rely on only one main algorithm. This makes it difficult for the system to overcome the shortcomings of various algorithms and cannot benefit from the combination of multiple recommendation algorithms. In addition, the widely used recommendation models are found to be unable to extract the finer features of users and movies, and the calculation time is very long. Meanwhile, the recommendation results are inaccurate, and they are not user-friendly. In this research, we design and construct a system to recommend movies which is consisted by a model based on a convolutional neural network consisting. By extracting the features of users and movies, we can calculate the direct similarity of different users' movies and then predict movie ratings to give recommendations. When extracting features of users and movies, we refer to traditional algorithms based on content and content. Through the "cosine similarity" and "user movie score matrix", the Item-Based Collaborative Filtering and User-Based Collaborative Filtering can be well implemented. To sum up, our movie recommendation system is based on Convolutional Neural Network (CNN) model and the performance of the system is improved by adopting multiple recommendation algorithms such as content-based recommendation and system filtering. We efficiently trained the neural network and finally built a movie recommendation system with a faster operation speed, although some parts are not perfect.

Detection and Correction of Abnormal IoT Data from Tea Plantations Based on Deep Learning

This paper proposes a data anomaly detection and correction algorithm for the tea plantation IoT system based on deep learning, aiming at the multi-cause and multi-feature characteristics of abnormal data. The algorithm is based on the Z-score standardization of the original data and the determination of sliding window size according to the sampling frequency. First, we construct a convolutional neural network (CNN) model to extract abnormal data. Second, based on the support vector machine (SVM) algorithm, the Gaussian radial basis function (RBF) and one-to-one (OVO) multiclassification method are used to classify the abnormal data. Then, after extracting the time points of abnormal data, a long short-term memory network is established for prediction with multifactor historical data. The predicted values are used to replace and correct the abnormal data. When multiple consecutive abnormal values are detected, a faulty sensor judgment is given, and the specific faulty sensor location is output. The results show that the accuracy rate and micro-specificity of abnormal data detection for the CNN-SVM model are 3–4% and 20–30% higher than those of the traditional CNN model, respectively. The anomaly detection and correction algorithm for tea plantation data established in this paper provides accurate performance.

A Multi-Task-Based Deep Multi-Scale Information Fusion Method for Intelligent Diagnosis of Bearing Faults

The use of deep learning for fault diagnosis is already a common approach. However, integrating discriminative information of fault types and scales into deep learning models for rich multitask fault feature diagnosis still deserves attention. In this study, a deep multitask-based multiscale feature fusion network model (MEAT) is proposed to address the limitations and poor adaptability of traditional convolutional neural network models for complex jobs. The model performed multidimensional feature extraction through convolution at different scales to obtain different levels of fault information, used a hierarchical attention mechanism to weight the fusion of features to achieve an accuracy of 99.95% for the total task of fault six classification, and considered two subtasks in fault classification to discriminate fault size and fault type through multi-task mapping decomposition. Of these, the highest accuracy of fault size classification reached 100%. In addition, Precision, ReCall, and Sacore F1 all reached the index of 1, which achieved the accurate diagnosis of bearing faults.

An efficient mobile model for insect image classification in the field pest management

Accurately recognizing insect pest in their larva phase is significant to take the early treatment on the infected crops, thus helping timely reduce the yield loss in agricultural products. The convolutional neural networks (CNNs)-based classification methods have become the most competitive methods to address many technical challenges related to image recognition in the field. Focusing on accurate and small models carried on mobile devices, this study proposed a novel pest classification method PCNet (Pest Classification Network) based on lightweight CNNs embedded attention mechanism. PCNet was designed with EfficientNet V2 as the backbone, and the coordinate attention mechanism (CA) was incorporated in this architecture to learn the inter-channel pest information and pest positional information of input images. Moreover, combining the feature maps output by mobile inverted bottleneck (MBConv) with the feature maps output by average pooling to develop the feature fusion module, which implements the feature fusion between shallow layers and deep layers to address the loss of insect pest features in the down-sampling procedures. In addition, a stochastic, pipeline-based data augmentation approach was adopted to randomly enhance data diversity and thus avoid model overfitting. The experimental results show that the PCNet model achieved recognition accuracy of 98.4 % on the self-built dataset consisting of 30 classes of larvae, which outperforms three classic CNN models (AlexNet, VGG16, and ResNet101), and four lightweight CNN models (ShuffleNet V2, MobileNet V3, EfficientNet V1 and V2). To further verify the robustness on different datasets, the proposed model was also tested on two other public datasets: IP102 and miniImageNet. The recognition accuracy of PCNet is 73.7 % on the IP102 dataset, outperforming other models and 94.0 % on miniImageNet dataset, which is only lower than that of ResNet101 and MobileNet V3. The number of PCNet parameters is 20.7 M, which is less than those of traditional classic CNN models. The satisfactory accuracy and small size of this model makes it suitable for real-time pest recognition in the field with resource constrained mobile devices. Our code will be available at https://github.com/pby521/PCNet/tree/master.

A Comparative Study of Different CNN Models and Transfer Learning Effect for Underwater Object Classification in Side-Scan Sonar Images

With the development of deep learning techniques, convolutional neural networks (CNN) are increasingly being used in image recognition for marine surveys and underwater object classification. Automatic recognition of targets on side-scan sonar (SSS) images using CNN can improve recognition accuracy and efficiency. However, the vast selection of CNN models makes it challenging to select models for target recognition in SSS images. Therefore, this paper aims to compare different CNN models’ prediction accuracy and computational performance comprehensively. First, four traditional CNN models were applied to train and predict the same submarine SSS dataset using both the original model and models with transfer learning methods. Then, we examined and studied the prediction accuracy and computation performance of four CNN models. Results showed that transfer learning enhances the accuracy of all CNN models, with lesser improvements for AlexNet and VGG-16 and greater improvements for GoogleNet and ResNet101. GoogleNet has the highest prediction of accuracy (100% in the train dataset and 94.27% in the test dataset) and good computational difficulty. The findings of this work are useful for future model selection in target recognition in SSS images.

GloWS-Net: A Deep Learning Framework for Retrieving Global Sea Surface Wind Speed Using Spaceborne GNSS-R Data

Spaceborne Global Navigation Satellite System Reflectometry (GNSS-R) is a new remote sensing technology that uses GNSS signals reflected from the Earth’s surface to estimate geophysical parameters. Because of its unique advantages such as high temporal and spatial resolutions, low observation cost, wide coverage and all-weather operation, it has been widely used in land and ocean remote sensing fields. Ocean wind monitoring is the main objective of the recently launched Cyclone GNSS (CYGNSS). In previous studies, wind speed was usually retrieved using features extracted from delay-Doppler maps (DDMs) and empirical geophysical model functions (GMFs). However, it is a challenge to employ the GMF method if using multiple sea state parameters as model input. Therefore, in this article, we propose an improved deep learning network framework to retrieve global sea surface wind speed using spaceborne GNSS-R data, named GloWS-Net. GloWS-Net considers the fusion of auxiliary information including ocean swell significant wave height (SWH), sea surface rainfall and wave direction to build an end-to-end wind speed retrieval model. In order to verify the improvement of the proposed model, ERA5 and Cross-Calibrated Multi-Platform (CCMP) wind data were used as reference for extensive testing to evaluate the wind speed retrieval performance of the GloWS-Net model and previous models (i.e., GMF, fully connected network (FCN) and convolutional neural network (CNN)). The results show that, when using ERA5 winds as ground truth, the root mean square error (RMSE) of the proposed GloWS-Net model is 23.98% better than that of the MVE method. Although the GloWS-Net model and the FCN model have similar RMSE (1.92 m/s), the mean absolute percentage error (MAPE) of the former is improved by 16.56%; when using CCMP winds as ground truth, the RMSE of the proposed GloWS-Net model is 2.16 m/s, which is 20.27% better than the MVE method. Compared with the FCN model, the MAPE is improved by 17.75%. Meanwhile, the GloWS-Net outperforms the FCN, traditional CNN, modified CNN (MCNN) and CyGNSSnet models in global wind speed retrieval especially at high wind speeds.

Rice Plaque Detection and Identification Based on an Improved Convolutional Neural Network

Rice diseases are extremely harmful to rice growth, and achieving the identification and rapid classification of rice disease spots is an essential means to promote intelligent rice production. However, due to the large variety of rice diseases and the similar appearance of some rice diseases, the existing deep learning methods are less effective at classification and detection. Aiming at such problems, this paper took the spot images of five common rice diseases as the research object and constructed a rice disease data set containing 2500 images of rice bacterial blight, sheath blight, flax leaf spot, leaf streak and rice blast, including 500 images of each disease. An improved lightweight deep learning network model was proposed to realize the accurate identification of disease types and disease spots. A rice disease image classification network was designed based on the RlpNet (rice leaf plaque net) network model, Which is the underlying network, in addition to the YOLOv3 target detection network model in order to achieve the optimization of the feature extraction link, i.e., upsampling by transposed convolution and downsampling by dilated convolution. The improved YOLOv3 model was compared with traditional convolutional neural network models, including the AlexNet, GoogLeNet, VGG-16 and ResNet-34 models, for disease recognition, and the results showed that the average recall, average precision, average F1-score and overall accuracy of the network model for rice disease classification were 91.84%, 92.14%, 91.87% and 91.84%, respectively, which were all improved compared with the traditional algorithms. The improved YOLOv3 network model was compared with FSSD, Faster-RCNN, YOLOv3 and YOLOv4 for spot detection studies, and the results showed that it could achieve a mean average precision (mAP) of 86.72%, a detection rate (DR) of 93.92%, a frames per second (FPS) rate of 63.4 and a false alarm rate (FAR) of only 5.12%. In summary, the comprehensive performance of the proposed model was better than that of the traditional YOLOv3 algorithm, so this study provides a new method for rice disease identification and disease spot detection. It also had good performance in terms of the common detection and classification of multiple rice diseases, which provides some support for the common differentiation of multiple rice diseases and has some practical application value.

Human and Machine Vision Based Indian Race Classification Using Modified-Convolutional Neural Network

The inter-class face classification problem is more reasonable than the intra-class classification problem. To address this issue, we have carried out empirical research on classifying Indian people to their geographical regions. This work aimed to construct a computational classification model for classifying Indian regional face images acquired from south and east regions of India, referring to human vision. We have created an Automated Human Intelligence System (AHIS) to evaluate human visual capabilities. Analysis of AHIS response showed that face shape is a discriminative feature among the other facial features. We have developed a modified convolutional neural network to characterize the human vision response to improve face classification accuracy. The proposed model achieved mean F1 and Matthew Correlation Coefficient (MCC) of 0.92 and 0.84, respectively, on the validation set, outperforming the traditional Convolutional Neural Network (CNN). The CNN-Contoured Face (CNN-FC) model is developed to train contoured face images to investigate the influence of face shape. Finally, to cross-validate the accuracy of these models, the traditional CNN model is trained on the same dataset. With an accuracy of 92.98%, the Modified-CNN (M-CNN) model has demonstrated that the proposed method could facilitate the tangible impact in intra-classification problems. A novel Indian regional face dataset is created for supporting this supervised classification work, and it will be available to the research community.

RBEBT: A ResNet-Based BA-ELM for Brain Tumor Classification

Brain tumor refers to the formation of abnormal cells in the brain. It can be divided into benign and malignant. The main diagnostic methods for brain tumors are plain X-ray film, Magnetic resonance imaging (MRI), and so on. However, these artificial diagnosis methods are easily affected by external factors. Scholars have made such impressive progress in brain tumors classification by using convolutional neural network (CNN). However, there are still some problems: (i) There are many parameters in CNN, which require much calculation. (ii) The brain tumor data sets are relatively small, which may lead to the overfitting problem in CNN. In this paper, our team proposes a novel model (RBEBT) for the automatic classification of brain tumors. We use fine-tuned ResNet18 to extract the features of brain tumor images. The RBEBT is different from the traditional CNN models in that the randomized neural network (RNN) is selected as the classifier. Meanwhile, our team selects the bat algorithm (BA) to optimize the parameters of RNN. We use five-fold cross-validation to verify the superiority of the RBEBT. The accuracy (ACC), specificity (SPE), precision (PRE), sensitivity (SEN), and F1-score (F1) are 99.00%, 95.00%, 99.00%, 100.00%, and 100.00%. The classification performance of the RBEBT is greater than 95%, which can prove that the RBEBT is an effective model to classify brain tumors.

Heterogenous-view occluded expression data recognition based on cycle-consistent adversarial network and K-SVD dictionary learning under intelligent cooperative robot environment

In space art design, the recognition of expression is of great help to the understanding of art. It is very difficult to obtain occlusion expression data from robot environment. In particular, it is very challenging to recognize the occluded expression. In the case of facial occlusion, it is difficult to extract the features of occluded expressions by traditional methods. In order to reduce the dependence of expression recognition on individuals, this paper proposes a cycle-consistent adversarial network and K-SVD dictionary learning method for occluded expression recognition in education management under robot environment. Firstly, the new method uses the cyclic-consistent generation adversarial network as the skeleton model, which can generate the un-occluded expression image without the need of paired data sets. Meanwhile, in order to improve the discriminant ability and image generation ability of the network, a multi-scale discriminator is used to construct the discriminant network. Then, the least squares and cyclic sensing loss are used to strengthen the constraints on the network model and improve the image quality. By subtracting the error matrix from the test sample, a clear image of the expression classification stage can be recovered. The clear image samples are decomposed into identity features and expression features by using the collaborative representation of two dictionaries. Finally, it is classified according to the contribution of each expression feature to the joint sparse representation. Experiments conducted on CK+, RAF-DB and SFEW datasets, the results show that the average accuracy of the new model is 98.44%, 87.12% and 62.17%, respectively. Compared with the traditional convolutional neural network models and advanced methods, this model effectively improves the accuracy of facial recognition in the case of facial occlusion.

Upper Limb Movement Decoding Scheme Based on Surface Electromyography Using Attention-Based Kalman Filter Scheme.

Convolutional neural network (CNN)-based models are widely used in human movement decoding based on surface electromyography. However, they capture only the spatial information of the surface electromyography and lack prior knowledge of the system, resulting in unsatisfactory decoding accuracy. To address these issues, we propose an attention-based Kalman filter scheme (AKFS), which uses an attention-based CNN model to better extract temporal information and a KF to add prior knowledge of the system. We further solve the problem of insufficient data due to the short training time of new subjects by using a transfer learning method based on a fine-tuning strategy. The proposed scheme was tested in four scenarios: intra-session, intra-session long-time use, inter-subject, and inter-subject with a fine-tuning strategy. The proposed attention-based CNN model outperformed the vanilla CNN model and a hybrid CNN-long short-term memory (LSTM) model in intra-session and intra-session long-time use. After fine-tuning with a small amount of data on a new subject, the attention-based CNN model achieved higher decoding accuracy than the vanilla CNN model and lower response time than CNN-LSTM model. Furthermore, the schemes with KF outperformed the schemes without KF in all scenarios. Our proposed scheme improves the decoding accuracy of the traditional CNN model for a single subject by better capturing the temporal information of the surface electromyography signal and increasing the prior knowledge of the system. Additionally, the proposed scheme can be easily transferred to a new subject using only a small amount of data.

Research on Fault Diagnosis Method Based on Improved CNN

Traditional fault diagnosis methods require complex signal processing and expert experience, and the accuracy of fault identification is low. To solve these problems, a fault diagnosis method based on an improved convolutional neural network (CNN) is proposed. Based on the traditional CNN model, a variety of convolution stride modes were added to extract features of different scales of signals and expand the feature dimension. Firstly, the vibration signals were collected and grouped. Then, the data were divided into a training set and a test set and input into improved CNN for feature extraction and model training to realize fault identification. The proposed model achieved a classification accuracy of 99.3% when testing the vibration data of the armored vehicle. Finally, the proposed model was used to classify different fault types of planetary gearboxes. The gradient-weighted class activation mapping (Grad-CAM) method was used to visualize the classification weight of samples. The results showed that the classification accuracy reaches 98% under various working conditions of the planetary gearbox.

Multimodal medical image fusion using convolutional neural network and extreme learning machine.

The emergence of multimodal medical imaging technology greatly increases the accuracy of clinical diagnosis and etiological analysis. Nevertheless, each medical imaging modal unavoidably has its own limitations, so the fusion of multimodal medical images may become an effective solution. In this paper, a novel fusion method on the multimodal medical images exploiting convolutional neural network (CNN) and extreme learning machine (ELM) is proposed. As a typical representative in deep learning, CNN has been gaining more and more popularity in the field of image processing. However, CNN often suffers from several drawbacks, such as high computational costs and intensive human interventions. To this end, the model of convolutional extreme learning machine (CELM) is constructed by incorporating ELM into the traditional CNN model. CELM serves as an important tool to extract and capture the features of the source images from a variety of different angles. The final fused image can be obtained by integrating the significant features together. Experimental results indicate that, the proposed method is not only helpful to enhance the accuracy of the lesion detection and localization, but also superior to the current state-of-the-art ones in terms of both subjective visual performance and objective criteria.

Ensembled Deep Learning for the Classification of Human Sperm Head Morphology

Infertility is a growing global health concern, with male factor infertility contributing to half of all cases. Semen analysis is crucial to infertility diagnostics. However, sperm morphology assessment, as a routine part of analysis, is still performed manually and is thus highly subjective. Here, a stacked ensemble of convolutional neural networks (CNNs) is presented for automated classification of human sperm head morphology. By combining traditional CNN models with modern residual and densely connected architectures using a multi‐class meta‐classifier, classification rate improvements of 2.7% (to 98.2%) and 2.3% (to 63.3%) on the HuSHeM and SCIAN‐MorphoSpermGS (SCIAN) datasets, respectively, are achieved. This considerable improvement in prediction performance is achieved as the meta‐classifier improves upon the individual classification rates of the base models by ≈8.5%. The ensembled deep learning model is a powerful step toward an automated sperm morphology analysis, providing new opportunities to standardize clinical practice and reduce treatment costs to improve patient treatment.

Two-Stream Mixed Convolutional Neural Network for American Sign Language Recognition.

The Convolutional Neural Network (CNN) has demonstrated excellent performance in image recognition and has brought new opportunities for sign language recognition. However, the features undergo many nonlinear transformations while performing the convolutional operation and the traditional CNN models are insufficient in dealing with the correlation between images. In American Sign Language (ASL) recognition, J and Z with moving gestures bring recognition challenges. This paper proposes a novel Two-Stream Mixed (TSM) method with feature extraction and fusion operation to improve the correlation of feature expression between two time-consecutive images for the dynamic gestures. The proposed TSM-CNN system is composed of preprocessing, the TSM block, and CNN classifiers. Two consecutive images in the dynamic gesture are used as inputs of streams, and resizing, transformation, and augmentation are carried out in the preprocessing stage. The fusion feature map obtained by addition and concatenation in the TSM block is used as inputs of the classifiers. Finally, a classifier classifies images. The TSM-CNN model with the highest performance scores depending on three concatenation methods is selected as the definitive recognition model for ASL recognition. We design 4 CNN models with TSM: TSM-LeNet, TSM-AlexNet, TSM-ResNet18, and TSM-ResNet50. The experimental results show that the CNN models with the TSM are better than models without TSM. The TSM-ResNet50 has the best accuracy of 97.57% for MNIST and ASL datasets and is able to be applied to a RGB image sensing system for hearing-impaired people.

A novel fault diagnosis method of rotating machinery via VMD, CWT and improved CNN

The rolling bearings play a vital role in mechanical production and transportation. However, when it appears abnormal, the fault characteristics are weak and different to be extracted in vibration signals which generally include pulse information reflecting fault type, independent vibration components caused by other normal mechanical components, noise in the surrounding environment, and so on.A hybrid method for rolling bearing fault diagnosis based on variational modal decomposition (VMD), continuous wavelet transform (CWT), convolutional neural network (CNN) and support vector machine (SVM) suitable for processing small samples is proposed. The pre-processed data are first subjected to VMD processing, and then the IMF reconstructed data are overlapped and sampled to obtain a two-dimensional time-frequency image using CWT. Then, a CNN model is constructed with selected hyperparameters, and image training samples are input into the CNN for model training. Finally, the pre-trained CNN model is selected to learn the test sample set layer by layer to realize the extraction of fault features and use SVM instead of Softmax function to identify and classify the faults.The effectiveness of the proposed method is verified with the Case Western Reserve University (CWRU) bearing vibration data and spindle device failure test bench dataset, where the classification accuracy of the former averaged 99.9% and that of the latter averaged 90.15%. The results validate the feasibility of the method and outperform traditional CNN models, feature extraction-based methods and other methods in terms of diagnostic accuracy.

A Time Series Transformer based method for the rotating machinery fault diagnosis

Fault diagnosis of rotating machinery is a significant engineering problem. In recent years, fault diagnosis methods have matured based on the Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). However, these traditional models have the problem of Long-Term Dependencies, leading to their feature extraction ability defect. To address these issues, we proposed a new method based on the Time Series Transformer (TST) to recognize the fault modes of the various rotating machinery. In this paper, firstly, we design a new tokens sequences generation method that can handle data in 1D format, namely time series tokenizer. Then the TST combining time series tokenizer and Transformer is presented. The test results on the given datasets show that the proposed method has better fault identification capability than traditional CNN and RNN models. Secondly, the effect of structural hyperparameters on fault diagnosis performance, computational complexity, and parameters number of the TST is analyzed in detail through experiments. The influence laws of some hyperparameters are obtained as well. Finally, the feature vectors in the embedding space are visualized via the t-Distributed Stochastic Neighbor Embedding (t-SNE) dimensionality reduction method. On this basis, the working pattern of TST is explained to a certain extent. Moreover, we find that the feature vectors extracted by the proposed method show the best intra-class compactness and inter-class separability compared with CNN and RNN models by analyzing their distribution form, which further demonstrates the effectiveness of the proposed method.