Weighted Loss Function Research Articles

A detection method for hybrid attacks in recommender systems

To defend recommender systems, some methods have been proposed to detect model-generative shilling attacks and group shilling attacks respectively. Unfortunately, these two categories of attacks are often mixed together to carry out actual attacks. Without the additional knowledge about attack categories, traditional detection methods are likely to be trapped in the poor performance under hybrids of model-generative shilling attacks and group shilling attacks. To simultaneously detect these hybrid attacks, we put forward a detector based on the graph convolutional networks (GCN). Firstly, we extract five user features from the item popularity sequence and rating values to characterize both model-generative shilling profiles and group shilling profiles. And we define the users’ distance to construct the user graph. Secondly, we develop a two-stage scheme for detecting shilling profiles based on user features and the user graph. In particular, we propose a cluster-based method to partially label user nodes, and then these labeled samples are fed into a GCN-based detector to training the model to identify the other tangled shilling profiles. In the GCN-based model, we present a weighted loss function with R-drop regularization to solve the over-fitting problem and the imbalanced classification problem for the specific detection task. Finally, we make extensive experiments on three datasets to evaluate the proposed detector. Experiential results demonstrate the efficacy of our method when detecting the hybrids of model-generative shilling attacks and group shilling attacks.

Multi-task learning on the edge for effective gender, age, ethnicity and emotion recognition

More and more real-world applications, especially in cognitive robotics, require the running of different computer vision algorithms in parallel on board of embedded devices with limited GPU and memory resources. Multi-task learning, namely the usage of the same model to perform multiple classification and/or regression tasks by learning a shared low level representation, revealed to be a valid solution to reduce computation and required memory space while preserving the accuracy. In this paper, we propose a solution for real-time user profiling based on a multi-task convolutional neural network (CNN) for gender, age, ethnicity and emotion recognition from face images. To find the best trade-off between accuracy and processing time, we evaluate three different architectures, specifically designed for the purpose, and backbones, based on MobileNet, ResNet and SENet, which include convolutional layers, residual blocks and attention modules that already demonstrated great potential in face analysis. We trained the multi-task neural network with a custom learning procedure, which solves the problems of missing labels, dataset imbalance and loss function imbalance through label masking, batch balancing and a custom weighted loss function; there are no other multi-task neural networks for face analysis that address all these challenges simultaneously.. The proposed solution demonstrated its effectiveness in the comparison with the corresponding single-task CNNs in terms of accuracy, processing time and memory space; in fact, the multi-task CNNs achieved a processing speed-up between 2.5 and 4 times and a reduction of the memory space between 2 and 4 times, while preserving the accuracy. Moreover, the useful insights that arise from the experiments allow to choose a solution for face analysis easily integrable into real applications on smart cameras and embedded systems and most suited for the specific application constraints in terms of computational resources.

Real-time single-channel speech enhancement based on causal attention mechanism

To achieve real-time single-channel speech enhancement, i.e., enhancing with no or low latency, this paper proposes a causal speech enhancement model with an attention mechanism based on Transformer. The model uses a causal codec with a U-net-like structure as the backbone network, which is improved with an upper triangle mask matrix and a single-side relative position representation on the basis of ensuring the causality. The mask matrix preserves the attentional focus on the historical global information and the single-side relative position representation focuses more on the information that needs attention in the local information. In addition, the weighted loss function in both time and frequency domains is used to guide the optimization direction of the training. Exhaustive comparison experiments are conducted on the Voice-Bank Demand dataset, and the experimental results show that the proposed causal model, compared with existing real-time single-channel speech enhancement models, not only possesses better enhancement results but also has faster training speed and fewer trainable parameters.

Joint estimation of PM2.5 and O3 over China using a knowledge-informed neural network

China has currently entered a critical stage of coordinated control of fine particulate matter (PM2.5) and ozone (O3), it is thus of tremendous value to accurately acquire high-resolution PM2.5 and O3 data. In contrast to traditional studies that usually separately estimate PM2.5 and O3, this study proposes a knowledge-informed neural network model for their joint estimation, in which satellite observations, reanalysis data, and ground station measurements are used. The neural network architecture is designed with the shared and specific inputs, the PM2.5-O3 interaction module, and the weighted loss function, which introduce the prior knowledge of PM2.5 and O3 into neural network modeling. Cross-validation (CV) results indicate that the inclusion of prior knowledge can improve the estimation accuracy, with R2 increasing from 0.872 to 0.911 and from 0.906 to 0.937 for PM2.5 and O3 estimation under sample-based CV, respectively. In addition, the proposed joint estimation model achieves comparable performance with the separate estimation model, but with higher efficiency. Mapping results of PM2.5 and O3 derived by the proposed model have demonstrated interesting findings in the spatial and temporal trends and variations over China.

An automatic method for microscopic diagnosis of diseases based on URCNN

Anthrax is a rare but dangerous disease for humans. A common way to diagnose this disease is the microscopic examination of slides containing tissue samples of patients. This paper aims to develop a reliable system in histopathological image analysis for the diagnosis of tissue diseases, metastasis, patient prognosis, etc. Automatic diagnosis of anthrax is investigated via the detection and segmentation of Bacillus anthracis bacteria and major immune system cells. Most recent models for instance segmentation are based on Mask-RCNN. It has an acceptable performance in most cases, but due to the challenges in the field of microscopic images, it fails to accurately detect and segment some of the important objects. Here, we have improved the performance of Mask-RCNN by making some modifications as follows: (1) A U-shaped structure is used as the mask branch in the head of Mask-RCNN that takes the advantage of combining multi-scale feature maps, (2) An enhanced FPN structure is proposed, which takes advantage of the squeeze and excitation-residual blocks and squeeze and excitation-inception blocks, (3) A hybrid weighted loss function composed of LBCE,LDiceandLIoU is proposed to update the weights and (4) Finally, a dropout layer is added after each FC layer in the classifier structure of the head Mask-RCNN architecture to improve the generalization power of the model. Experimental results show that the proposed model outperforms the state-of-the-art architectures and is a reliable system for the automatic diagnosis of anthrax.

The relationship between weight loss and cognitive function in bariatric surgery.

Previously, we reported short-term improvements in auditory attention, oromotor processing speed, and executive function during the active weight loss phase following bariatric surgery that persisted out to 3months. In this study, our aims were to investigate the relationship between weight loss and cognitive performance in these patients 1 year following vertical sleeve gastrectomy (VSG) and Roux-en Y gastric bypass (RYGB) surgery and to determine whether preoperative cognitive performance predicted weight loss. Adult women ages 18-55 approved for bariatric surgery completed a cognitive battery prior to and at 2, 12, 24, and 52weeks following VSG (N = 17) or RYGB (N = 18). Scores from each task were assigned to one of the following cognitive domains: auditory attention, processing speed, memory, and executive functioning. Weight loss and cognitive scores for each domain were calculated and compared between cohorts. RYGB surgery resulted in greater weight loss at 1-year follow-up relative to VSG. Both VSG and RYGB procedures resulted in improved performance on different measures of auditory attention and both surgery groups improved across all processing speed tasks. Within the executive function domain, both groups showed improvements, but only the RYGB procedure resulted in improved performance in the Trail Making Test. Baseline auditory attention and memory performance predicted weight loss at 1 year following RYGB but not VSG surgery. Controlling for baseline cognitiveperformance, percent total weight loss predicted auditory attention at 1 year following RYGB but not VSG surgery. Bariatric surgery type may result in selective improvements in cognition during the first year following surgery. Presurgical cognitive performance as well as surgery type appears to influence weight loss outcomes.

Robust Extreme Learning Machine Using New Activation and Loss Functions Based on M-Estimation for Regression and Classification

This paper provides an analysis of the combining effect of novel activation function and loss function based on M-estimation in application to extreme learning machine (ELM), a feed-forward neural network. Due to the computational efficiency and classification/prediction accuracy of ELM and its variants, they have been widely exploited in the development of new technologies and applications. However, in real applications, the performance of classical ELMs deteriorates in the presence of outliers, thus, negatively impacting the precision and accuracy of the system. To further enhance the performance of ELM and its variants, we proposed novel activation functions based on the psi function of M and redescend the M-estimation method along with the smooth l 2-norm weight-loss functions to reduce the negative impact of the outliers. The proposed psi functions of several M and redescending M-estimation methods are more flexible to make more distinct features space. For the first time, the idea of the psi function as an activation function in the neural network is introduced in the literature to ensure accurate prediction. In addition, new robust l 2 norm-loss functions based on M and redescending M-estimation are proposed to deal with outliers efficiently in ELM. To evaluate the performance of the proposed methodology against other state-of-the-art techniques, experiments have been performed in diverse environments, which show promising improvements in application to regression and classification problems.

Asynchronous functional linear regression models for longitudinal data in reproducing kernel Hilbert space.

Motivated by the analysis of longitudinal neuroimaging studies, we study the longitudinal functional linear regression model under asynchronous data setting for modeling the association between clinical outcomes and functional (or imaging) covariates. In the asynchronous data setting, both covariates and responses may be measured at irregular and mismatched time points, posing methodological challenges to existing statistical methods. We develop a kernel weighted loss function with roughness penalty to obtain the functional estimator and derive its representer theorem. The rate of convergence, a Bahadur representation, and the asymptotic pointwise distribution of the functional estimator are obtained under the reproducing kernel Hilbert space framework. We propose a penalized likelihood ratio test to test the nullity of the functional coefficient, derive its asymptotic distribution under the null hypothesis, and investigate the separation rate under the alternative hypotheses. Simulation studies are conducted to examine the finite-sample performance of the proposed procedure. We apply the proposed methods to the analysis of multitype data obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study, which reveals significant association between 21 regional brain volume density curves and the cognitive function. Data used in preparation of this paper were obtained from the ADNI database (adni.loni.usc.edu).

Ultra-Short-Term Load Dynamic Forecasting Method Considering Abnormal Data Reconstruction Based on Model Incremental Training

In order to reduce the influence of abnormal data on load forecasting effects and further improve the training efficiency of forecasting models when adding new samples to historical data set, an ultra-short-term load dynamic forecasting method considering abnormal data reconstruction based on model incremental training is proposed in this paper. Firstly, aiming at the abnormal data in ultra-short-term load forecasting, a load abnormal data processing method based on isolation forests and conditional adversarial generative network (IF-CGAN) is proposed. The isolation forest algorithm is used to accurately eliminate the abnormal data points, and a conditional generative adversarial network (CGAN) is constructed to interpolate the abnormal points. The load-influencing factors are taken as the condition constraints of the CGAN, and the weighted loss function is introduced to improve the reconstruction accuracy of abnormal data. Secondly, aiming at the problem of low model training efficiency caused by the new samples in the historical data set, a model incremental training method based on a bidirectional long short-term memory network (Bi-LSTM) is proposed. The historical data are used to train the Bi-LSTM, and the transfer learning is introduced to process the incremental data set to realize the adaptive and rapid adjustment of the model weight and improve the model training efficiency. Finally, the real power grid load data of a region in eastern China are used for simulation analysis. The calculation results show that the proposed method can reconstruct the abnormal data more accurately and improve the accuracy and efficiency of ultra-short-term load forecasting.

Automatic Defect Detection of Pavement Diseases

Pavement disease detection is an important task for ensuring road safety. Manual visual detection requires a significant amount of time and effort. Therefore, an automated road disease identification technique is required to guarantee that city tasks are performed. However, due to the irregular shape and large-scale differences in road diseases, as well as the imbalance between the foreground and background, the task is challenging. Because of this, we created the deep convolution neural network—DASNet, which can be used to identify road diseases automatically. The network employs deformable convolution instead of regular convolution as the feature pyramid’s input, adds the same supervision signal to the multi-scale features before feature fusion, decreases the semantic difference, extracts context information by residual feature enhancement, and reduces the information loss of the pyramid’s top-level feature map. Considering the unique shape of road diseases, imbalance problems between the foreground and background are common, therefore, we introduce the sample weighted loss function. In order to prove the superiority and effectiveness of this method, it is compared to the latest method. A large number of experiments show that this method is superior in accuracy to other methods, specifically, under the COCO evaluation metric, compared with the Faster RCNN baseline, the proposed method obtains a 41.1 mAP and 3.4 AP improvement.

A coarse-to-fine cascade deep learning neural network for segmenting cerebral aneurysms in time-of-flight magnetic resonance angiography.

BackgroundAccurate segmentation of unruptured cerebral aneurysms (UCAs) is essential to treatment planning and rupture risk assessment. Currently, three-dimensional time-of-flight magnetic resonance angiography (3D TOF-MRA) has been the most commonly used method for screening aneurysms due to its noninvasiveness. The methods based on deep learning technologies can assist radiologists in achieving accurate and reliable analysis of the size and shape of aneurysms, which may be helpful in rupture risk prediction models. However, the existing methods did not accomplish accurate segmentation of cerebral aneurysms in 3D TOF-MRA.MethodsThis paper proposed a CCDU-Net for segmenting UCAs of 3D TOF-MRA images. The CCDU-Net was a cascade of a convolutional neural network for coarse segmentation and the proposed DU-Net for fine segmentation. Especially, the dual-channel inputs of DU-Net were composed of the vessel image and its contour image which can augment the vascular morphological information. Furthermore, a newly designed weighted loss function was used in the training process of DU-Net to promote the segmentation performance.ResultsA total of 270 patients with UCAs were enrolled in this study. The images were divided into the training (N = 174), validation (N = 43), and testing (N = 53) cohorts. The CCDU-Net achieved a dice similarity coefficient (DSC) of 0.616 ± 0.167, Hausdorff distance (HD) of 5.686 ± 7.020 mm, and volumetric similarity (VS) of 0.752 ± 0.226 in the testing cohort. Compared with the existing best method, the DSC and VS increased by 18% and 5%, respectively, while the HD decreased by one-tenth.ConclusionsWe proposed a CCDU-Net for segmenting UCAs in 3D TOF-MRA, and the obtained results show that the proposed method outperformed other existing methods.

Clustering-based spatial transfer learning for short-term ozone forecasting

Ground-level ozone is a critical atmospheric pollutant, and high concentrations of ozone can damage human health, affect plant growth and cause ecological harm. Traditional chemical transport models and popular machine learning models have difficulty in predicting ozone concentrations, especially in times with high concentrations. We proposes a clustering-based spatial transfer learning Multilayer Perceptron (SPTL-MLP) to predict ozone concentration at the target observation station for the next three days. We use k-means clustering algorithm to find similar stations and train them together to get a base model for spatial transfer learning. For practical applications, a weighted loss function has been designed with an extra emphasis on reducing prediction errors of high ozone concentrations. Evaluation using historical data of stations in Germany shows that our SPTL-MLP model has a smaller error (reduced by 9.13%) and higher prediction accuracies of ozone exceedances (improved by 8.21% and 16.9%) compared to MLP (without spatial transfer). The results demonstrate the effectiveness of the SPTL-MLP in the short-term ozone forecast. It can be used for timely warning of ozone exceedances and help governments to detect air quality.

ClarifyNet: A high-pass and low-pass filtering based CNN for single image dehazing

Dehazing refers to removing the haze and restoring the details from hazy images. In this paper, we propose ClarifyNet, a novel, end-to-end trainable, convolutional neural network architecture for single image dehazing. We note that a high-pass filter detects sharp edges, texture, and other fine details in the image, whereas a low-pass filter detects color and contrast information. Based on this observation, our key idea is to train ClarifyNet on ground-truth haze-free images, low-pass filtered images, and high-pass filtered images. Based on this observation, we present a shared-encoder multi-decoder model ClarifyNet which employs interconnected parallelization. While training, ground-truth haze-free images, low-pass filtered images, and high-pass filtered images undergo multi-stage filter fusion and attention. By utilizing a weighted loss function composed of SSIM loss and L1 loss, we extract and propagate complementary features. We comprehensively evaluate ClarifyNet on I-HAZE, O-HAZE, Dense-Haze, NH-HAZE, SOTS-Indoor, SOTS-Outdoor, HSTS, and Middlebury datasets. We use PSNR and SSIM metrics and compare the results with previous works. For most datasets, ClarifyNet provides the highest scores. On using EfficientNet-B6 as the backbone, ClarifyNet has 18 M parameters (model size of ∼71 MB) and a throughput of 8 frames-per-second while processing images of size 2048 × 1024.

Effects of a whole food diet on immune function and inflammatory phenotype in healthy dogs: A randomized, open-labeled, cross-over clinical trial.

Whole foods in humans decrease inflammation and risk for various diseases, as well as increase weight loss and immune function. Nutrition has been shown to be an integral component in the management of various diseases in dogs but the immunologic and anti-inflammatory effects of whole food diets have not been explored. Therefore, our objective was to assess the effect of feeding a whole food diet on immune function and inflammatory phenotype in healthy dogs. A prospective, randomized, open-labeled, cross-over clinical trial was performed. Sixteen healthy client-owned dogs were fed either a whole food or an extruded dry diet, and after 67 days, they were fed the alternate diet for an additional 67 days. Blood samples were obtained at the completion of each treatment arm (i.e., days 67 and 134). Serum c-reactive protein (CRP), haptoglobin (Hp), and serum amyloid-A (SAA) were measured with ELISA assays. Whole blood cultures were performed with exposure to a phosphate-buffered solution (PBS), lipopolysaccharide (LPS), and lipoteichoic acid (LTA). A canine specific multiplex bead-based assay was then used to measure tumor necrosis factor (TNF)-α, interleukin (IL)-6, IL-10, granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-2, IL-8, and monocyte chemoattractant protein (MCP)-1 concentrations. Granulocyte/monocyte (GM) phagocytosis and oxidative burst associated with Escherichia coli were evaluated via flow cytometry. Dogs fed a whole food diet had significantly lower TNF-α-to-IL-10 ratios (P = 0.05) and higher production of IL-8 (P = 0.03) with LTA-exposed leukocytes compared to dogs fed an extruded dry diet. There were no between-treatment differences in the remaining leukocyte cytokine responses, serum CRP, Hp, SAA concentrations, or GM phagocytic and oxidative burst capacities. Whole food diets could have immunomodulatory effects in dogs. Future studies in non-healthy dogs are warranted.

Deep Learning-Based Localization with Urban Electromagnetic and Geographic Information

There is a growing demand for localization of illegal signal sources, aiming to guarantee the security of urban electromagnetic environment. The performance of traditional localization methods is limited due to the non-line-of-sight (NLOS) propagation and sparse layouts of sensors. In this paper, a deep learning-based localization method is proposed to overcome these issues in urban scenarios. Firstly, a model of electromagnetic wave propagation considered with geographic information is proposed to prepare reliable datasets for intelligent cognition of urban electromagnetic environment. Then, this paper improves an hourglass neural network which consists of downsampling and upsampling layers to learn the propagation features from sensing data. The core modules of VGG and ResNet are, respectively, utilized as feature extractors in downsampling. Moreover, this paper proposes a weighted loss function to expand the attention on position features, in order to improve the performance of localization with sparse layouts of sensors. Representative numerical results are discussed to assess the proposed method. ResNet-based extractor performs more efficiently than VGG-based extractor, and the proposed weighted loss function increases the localization accuracy by more than 50%. Additionally, the established geographic model supports qualitative and quantitative evaluation of the robustness with varied degree of NLOS propagation. Compared with other deep learning-based algorithms, the proposed method presents the more robust and superior performance under severe NLOS propagation and sparse sensing conditions.

Resampling-based cost loss attention network for explainable imbalanced diabetic retinopathy grading

Diabetic retinopathy (DR) is currently considered to be one of the most common diseases that cause blindness. However, DR grading methods are still challenged by the presence of imbalanced class distributions, small lesions, low accuracy of small sample classes and poor explainability. To address these issues, a resampling-based cost loss attention network for explainable imbalanced diabetic retinopathy grading is proposed. First, the progressively-balanced resampling strategy is put forward to create a balanced training data by mixing the two sets of samples obtained from instance-based sampling and class-based sampling. Subsequently, a neuron and normalized channel-spatial attention module (Neu-NCSAM) is designed to learn the global features with 3-D weights and a weight sparsity penalty is applied to the attention module to suppress irrelevant channels or pixels, thereby capturing detailed small lesion information. Thereafter, a weighted loss function of the Cost-Sensitive (CS) regularization and Gaussian label smoothing loss, called cost loss, is proposed to intelligently penalize the incorrect predictions and thus to improve the grading accuracy of small sample classes. Finally, the Gradient-weighted Class Activation Mapping (Grad-CAM) is performed to acquire the localization map of the questionable lesions in order to visually interpret and understand the effect of our model. Comprehensive experiments are carried out on two public datasets, and the subjective and objective results demonstrate that the proposed network outperforms the state-of-the-art methods and achieves the best DR grading results with 83.46%, 60.44%, 65.18%, 63.69% and 92.26% for Kappa, BACC, MCC, F1 and mAUC, respectively.

Vehicle trajectory prediction considering aleatoric uncertainty

Trajectory prediction is imperative in the operation of autonomous vehicles because it aids in understanding the surrounding environment through perception fusion of multiple sensors and high-accuracy localisation. Furthermore, the output of algorithms used for trajectory prediction is used in decision making and path planning. However, the accuracy of trajectory prediction is dependent on the perception system particularly when low-cost sensors are used or severe weather conditions are encountered. Data-dependent and task-dependent aleatoric uncertainty affects the trajectory prediction task. Therefore, in this study, we proposed a network for trajectory prediction that integrates the temporal pattern attention and the graph convolutional sequence encoding to address the data-dependent uncertainty caused by observation noise and machine-perceptual errors. Furthermore, an adaptive weight loss function was developed to address the training difficulties caused by loss-dominated tasks (task-dependent aleatoric uncertainty) in a trajectory prediction network. The results of our experiment indicate a 20% improvement in predictive performance over existing methods under a 5-s prediction horizon. Furthermore, an experiment comparing robustness to heteroscedastic uncertainty confirmed that the proposed method improves trajectory prediction robustness against observation noise and perceptual errors.

Grid adaptive reduced-order model of fluid flow based on graph convolutional neural network

In the interdisciplinary field of data-driven models and computational fluid mechanics, the reduced-order model for flow field prediction is mainly constructed by a convolutional neural network (CNN) in recent years. However, the standard CNN is only applicable to data with Euclidean spatial structure, while data with non-Euclidean properties can only be convolved after pixelization, which usually leads to decreased data accuracy. In this work, a novel data-driven framework based on graph convolution network (GCN) is proposed to allow the convolution operator to predict fluid dynamics on non-uniform structured or unstructured mesh data. This is achieved by the fact that the graph data inherit the spatial characteristics of the mesh and by the message passing mechanism of GCN. The conversion method from the form of mesh data to graph data and the operation mechanism of GCN are clarified. Moreover, additional relevance features and weight loss function of the dataset are also investigated to improve the model performance. The model learns an end-to-end mapping between the mesh spatial features and the physical flow field. Through our studies of various cases of internal flow, it is shown that the proposed GCN-based model offers excellent adaptability to non-uniformly distributed mesh data, while also achieving a high accuracy and three-order speedup compared with numerical simulation. Our framework generalizes the graph convolution network to flow field prediction and opens the door to further extending GCN to most existing data-driven architectures of fluid dynamics in the future.

Cloudformer V2: Set Prior Prediction and Binary Mask Weighted Network for Cloud Detection

Cloud detection is an essential step in optical remote sensing data processing. With the development of deep learning technology, cloud detection methods have made remarkable progress. Among them, researchers have started to try to introduce Transformer into cloud detection tasks due to its excellent performance in image semantic segmentation tasks. However, the current Transformer-based methods suffer from training difficulty and low detection accuracy of small clouds. To solve these problems, this paper proposes Cloudformer V2 based on the previously proposed Cloudformer. For the training difficulty, Cloudformer V2 uses Set Attention Block to extract intermediate features as Set Prior Prediction to participate in supervision, which enables the model to converge faster. For the detection of small clouds, Cloudformer V2 decodes the features by a multi-scale Transformer decoder, which uses multi-resolution features to improve the modeling accuracy. In addition, a binary mask weighted loss function (BW Loss) is designed to construct weights by counting pixels classified as clouds; thus, guiding the network to focus on features of small clouds and improving the overall detection accuracy. Cloudformer V2 is experimented on the dataset from GF-1 satellite and has excellent performance.

Predicting the dynamic process and model parameters of vector optical solitons under coupled higher-order effects via WL-tsPINN

We propose the two-subnet physical information neural network with the weighted loss function (WL-tsPINN) to study the higher-order effects of ultra-short pulses in birefringence fiber transmission and analyze the formation mechanism of vector solitons. We predict the dynamical process of mixed-type single/double soliton and soliton molecules based on the higher-order coupled nonlinear Schrödinger equation (CNLSE) by this WL-tsPINN method. Moreover, we deduce the physical coefficients of the higher-order CNLSE from the mixed single soliton solution. Deep learning based on neural network is a powerful tool for further study of higher-order CNLSE and has potential significance for further study of soliton dynamics.