Daily scale air quality index forecasting using bidirectional recurrent neural networks: Case study of Delhi, India

This research was established to accurately forecast daily scale air quality index (AQI) which is an essential environmental index for decision-making. Researchers have projected different types of models and methodologies for AQI forecasting, such as statistical techniques, machine learning (ML), and most recently deep learning (DL) models. The modelling development was adopted for Delhi city, India which is a major city with air pollution issues simialir to entire urban cities of India especially during winter seasons. This research was predicted AQI using different versions of DL models including Long-Short Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM) and Bidirectional Recurrent Neural Networks (Bi-RNN) in addition to Kernel Ridge Regression (KRR). Results indicated that Bi-RNN model consistently outperformed the other models in both training and testing phases, while the KRR model consistently displayed the weakest performance. The outstanding performance of the models development displayed the requirement of adequate data to train the models. The outcomes of the models showed that LSTM, BI-LSTM, KRR had lower performance compared with Bi-RNN models. Statistically, Bi-RNN model attained maximum cofficient of determination (R2 = 0.954) and minimum root mean square error (RMSE = 25.755). The proposed model in this research revealed the robust predictable to provide a valuable base for decision-making in the expansion of combined air pollution anticipation and control policies targeted at addressing composite air pollution problems in the Delhi city.

Quantitative Evaluation of the Pore and Window Sizes of Tissue Engineering Scaffolds on Scanning Electron Microscope Images Using Deep Learning

The morphological characteristics of tissue engineering scaffolds, such as pore and window diameters, are crucial, as they directly impact cell-material interactions, attachment, spreading, infiltration of the cells, degradation rate and the mechanical properties of the scaffolds. Scanning electron microscopy (SEM) is one of the most commonly used techniques for characterizing the microarchitecture of tissue engineering scaffolds due to its advantages, such as being easily accessible and having a short examination time. However, SEM images provide qualitative data that need to be manually measured using software such as ImageJ to quantify the morphological features of the scaffolds. As it is not practical to measure each pore/window in the SEM images as it requires extensive time and effort, only the number of pores/windows is measured and assumed to represent the whole sample, which may cause user bias. Additionally, depending on the number of samples and groups, a study may require measuring thousands of samples and the human error rate may increase. To overcome such problems, in this study, a deep learning model (Pore D2) was developed to quantify the morphological features (such as the pore size and window size) of the open-porous scaffolds automatically for the first time. The developed algorithm was tested on emulsion-templated scaffolds fabricated under different fabrication conditions, such as changing mixing speed, temperature, and surfactant concentration, which resulted in scaffolds with various morphologies. Along with the developed model, blind manual measurements were taken, and the results showed that the developed tool is capable of quantifying pore and window sizes with a high accuracy. Quantifying the morphological features of scaffolds fabricated under different circumstances and controlling these features enable us to engineer tissue engineering scaffolds precisely for specific applications. Pore D2, an open-source software, is available for everyone at the following link: https://github.com/ilaydakaraca/PoreD2.

PCA and logistic regression in 2-[18F]FDG PET neuroimaging as an interpretable and diagnostic tool for Alzheimer's disease

Objective.to develop an optimization and training pipeline for a classification model based on principal component analysis and logistic regression using neuroimages from PET with 2-[18F]fluoro-2-deoxy-D-glucose (FDG PET) for the diagnosis of Alzheimer's disease (AD).Approach.as training data, 200 FDG PET neuroimages were used, 100 from the group of patients with AD and 100 from the group of cognitively normal subjects (CN), downloaded from the repository of the Alzheimer's Disease Neuroimaging Initiative (ADNI). Regularization methods L1 and L2 were tested and their respective strength varied by the hyperparameter C. Once the best combination of hyperparameters was determined, it was used to train the final classification model, which was then applied to test data, consisting of 192 FDG PET neuroimages, 100 from subjects with no evidence of AD (nAD) and 92 from the AD group, obtained at the Centro de Diagnóstico por Imagem (CDI).Main results.the best combination of hyperparameters was L1 regularization andC≈ 0.316. The final results on test data were accuracy = 88.54%, recall = 90.22%, precision = 86.46% and AUC = 94.75%, indicating that there was a good generalization to neuroimages outside the training set. Adjusting each principal component by its respective weight, an interpretable image was obtained that represents the regions of greater or lesser probability for AD given high voxel intensities. The resulting image matches what is expected by the pathophysiology of AD.Significance.our classification model was trained on publicly available and robust data and tested, with good results, on clinical routine data. Our study shows that it serves as a powerful and interpretable tool capable of assisting in the diagnosis of AD in the possession of FDG PET neuroimages. The relationship between classification model output scores and AD progression can and should be explored in future studies.

An MRI brain tumor segmentation method based on improved U-Net

In order to improve the segmentation effect of brain tumor images and address the issue of feature information loss during convolutional neural network (CNN) training, we present an MRI brain tumor segmentation method that leverages an enhanced U-Net architecture. First, the ResNet50 network was used as the backbone network of the improved U-Net, the deeper CNN can improve the feature extraction effect. Next, the Residual Module was enhanced by incorporating the Convolutional Block Attention Module (CBAM). To increase characterization capabilities, focus on important features and suppress unnecessary features. Finally, the cross-entropy loss function and the Dice similarity coefficient are mixed to compose the loss function of the network. To solve the class unbalance problem of the data and enhance the tumor area segmentation outcome. The method's segmentation performance was evaluated using the test set. In this test set, the enhanced U-Net achieved an average Intersection over Union (IoU) of 86.64% and a Dice evaluation score of 87.47%. These values were 3.13% and 2.06% higher, respectively, compared to the original U-Net and R-Unet models. Consequently, the proposed enhanced U-Net in this study significantly improves the brain tumor segmentation efficacy, offering valuable technical support for MRI diagnosis and treatment.

A Pipeline Defect Instance Segmentation System Based on SparseInst

Deep learning algorithms have achieved encouraging results for pipeline defect segmentation. However, existing defect segmentation methods may encounter challenges in accurately segmenting the complex features of pipeline defects and suffer from low processing speeds. Therefore, in this study, we propose Pipe-Sparse-Net, a pipeline defect segmentation system that combines StyleGAN3 to segment the complex forms of underground drainage pipe defects. First, we introduce a data augmentation algorithm based on StyleGAN3 to enlarge the dataset. Next, we propose Pipe-Sparse-Net, a pipeline segmentation model based on SparseInst, to accurately predict the defect regions in drainage pipes. Experimental results demonstrate that the segmentation accuracy of this model can reach 91.4% with a processing speed of 56.7 frames per second (FPS). To validate the superiority of this method, comparative experiments were conducted against Yolact, Condinst, and Mask R-CNN, and the model achieved a speed improvement of 45% while increasing the accuracy by more than 4%.

Detection of Corneal Ulcer Using a Genetic Algorithm-Based Image Selection and Residual Neural Network

Corneal ulcer is one of the most devastating eye diseases causing permanent damage. There exist limited soft techniques available for detecting this disease. In recent years, deep neural networks (DNN) have significantly solved numerous classification problems. However, many samples are needed to obtain reasonable classification performance using a DNN with a huge amount of layers and weights. Since collecting a data set with a large number of samples is usually a difficult and time-consuming process, very large-scale pre-trained DNNs, such as the AlexNet, the ResNet and the DenseNet, can be adapted to classify a dataset with a small number of samples, through the utility of transfer learning techniques. Although such pre-trained DNNs produce successful results in some cases, their classification performances can be low due to many parameters, weights and the emergence of redundancy features that repeat themselves in many layers in som cases. The proposed technique removes these unnecessary features by systematically selecting images in the layers using a genetic algorithm (GA). The proposed method has been tested on ResNet on a small-scale dataset which classifies corneal ulcers. According to the results, the proposed method significantly increased the classification performance compared to the classical approaches.

Artificial intelligence in thyroid ultrasound

Artificial intelligence (AI), particularly deep learning (DL) algorithms, has demonstrated remarkable progress in image-recognition tasks, enabling the automatic quantitative assessment of complex medical images with increased accuracy and efficiency. AI is widely used and is becoming increasingly popular in the field of ultrasound. The rising incidence of thyroid cancer and the workload of physicians have driven the need to utilize AI to efficiently process thyroid ultrasound images. Therefore, leveraging AI in thyroid cancer ultrasound screening and diagnosis cannot only help radiologists achieve more accurate and efficient imaging diagnosis but also reduce their workload. In this paper, we aim to present a comprehensive overview of the technical knowledge of AI with a focus on traditional machine learning (ML) algorithms and DL algorithms. We will also discuss their clinical applications in the ultrasound imaging of thyroid diseases, particularly in differentiating between benign and malignant nodules and predicting cervical lymph node metastasis in thyroid cancer. Finally, we will conclude that AI technology holds great promise for improving the accuracy of thyroid disease ultrasound diagnosis and discuss the potential prospects of AI in this field.

Research on two-class and four-class action recognition based on EEG signals

BMI has attracted widespread attention in the past decade, which has greatly improved the living conditions of patients with motor disorders. The application of EEG signals in lower limb rehabilitation robots and human exoskeleton has also been gradually applied by researchers. Therefore, the recognition of EEG signals is of great significance. In this paper, a CNN-LSTM neural network model is designed to study the two-class and four-class motion recognition of EEG signals. In this paper, a brain-computer interface experimental scheme is designed. Combining the characteristics of EEG signals, the time-frequency characteristics of EEG signals and event-related potential phenomena are analyzed, and the ERD/ERS characteristics are obtained. Pre-process EEG signals, and propose a CNN-LSTM neural network model to classify the collected binary and four-class EEG signals. The experimental results show that the CNN-LSTM neural network model has a good effect, and its average accuracy and kappa coefficient are higher than the other two classification algorithms, which also shows that the classification algorithm selected in this paper has a good classification effect.

Estimation of Off-Target Dicamba Damage on Soybean Using UAV Imagery and Deep Learning

Weeds can cause significant yield losses and will continue to be a problem for agricultural production due to climate change. Dicamba is widely used to control weeds in monocot crops, especially genetically engineered dicamba-tolerant (DT) dicot crops, such as soybean and cotton, which has resulted in severe off-target dicamba exposure and substantial yield losses to non-tolerant crops. There is a strong demand for non-genetically engineered DT soybeans through conventional breeding selection. Public breeding programs have identified genetic resources that confer greater tolerance to off-target dicamba damage in soybeans. Efficient and high throughput phenotyping tools can facilitate the collection of a large number of accurate crop traits to improve the breeding efficiency. This study aimed to evaluate unmanned aerial vehicle (UAV) imagery and deep-learning-based data analytic methods to quantify off-target dicamba damage in genetically diverse soybean genotypes. In this research, a total of 463 soybean genotypes were planted in five different fields (different soil types) with prolonged exposure to off-target dicamba in 2020 and 2021. Crop damage due to off-target dicamba was assessed by breeders using a 1-5 scale with a 0.5 increment, which was further classified into three classes, i.e., susceptible (≥3.5), moderate (2.0 to 3.0), and tolerant (≤1.5). A UAV platform equipped with a red-green-blue (RGB) camera was used to collect images on the same days. Collected images were stitched to generate orthomosaic images for each field, and soybean plots were manually segmented from the orthomosaic images. Deep learning models, including dense convolutional neural network-121 (DenseNet121), residual neural network-50 (ResNet50), visual geometry group-16 (VGG16), and Depthwise Separable Convolutions (Xception), were developed to quantify crop damage levels. Results show that the DenseNet121 had the best performance in classifying damage with an accuracy of 82%. The 95% binomial proportion confidence interval showed a range of accuracy from 79% to 84% (p-value ≤ 0.01). In addition, no extreme misclassifications (i.e., misclassification between tolerant and susceptible soybeans) were observed. The results are promising since soybean breeding programs typically aim to identify those genotypes with 'extreme' phenotypes (e.g., the top 10% of highly tolerant genotypes). This study demonstrates that UAV imagery and deep learning have great potential to high-throughput quantify soybean damage due to off-target dicamba and improve the efficiency of crop breeding programs in selecting soybean genotypes with desired traits.

An Insight of Deep Learning Based Demand Forecasting in Smart Grids

Smart grids are able to forecast customers' consumption patterns, i.e., their energy demand, and consequently electricity can be transmitted after taking into account the expected demand. To face today's demand forecasting challenges, where the data generated by smart grids is huge, modern data-driven techniques need to be used. In this scenario, Deep Learning models are a good alternative to learn patterns from customer data and then forecast demand for different forecasting horizons. Among the commonly used Artificial Neural Networks, Long Short-Term Memory networks-based on Recurrent Neural Networks-are playing a prominent role. This paper provides an insight into the importance of the demand forecasting issue, and other related factors, in the context of smart grids, and collects some experiences of the use of Deep Learning techniques, for demand forecasting purposes. To have an efficient power system, a balance between supply and demand is necessary. Therefore, industry stakeholders and researchers should make a special effort in load forecasting, especially in the short term, which is critical for demand response.

An automatic music generation and evaluation method based on transfer learning

In recent years, deep learning has seen remarkable progress in many fields, especially with many excellent pre-training models emerged in Natural Language Processing(NLP). However, these pre-training models can not be used directly in music generation tasks due to the different representations between music symbols and text. Compared with the traditional presentation method of music melody that only includes the pitch relationship between single notes, the text-like representation method proposed in this paper contains more melody information, including pitch, rhythm and pauses, which expresses the melody in a form similar to text and makes it possible to use existing pre-training models in symbolic melody generation. In this paper, based on the generative pre-training-2(GPT-2) text generation model and transfer learning we propose MT-GPT-2(music textual GPT-2) model that is used in music melody generation. Then, a symbolic music evaluation method(MEM) is proposed through the combination of mathematical statistics, music theory knowledge and signal processing methods, which is more objective than the manual evaluation method. Based on this evaluation method and music theories, the music generation model in this paper are compared with other models (such as long short-term memory (LSTM) model,Leak-GAN model and Music SketchNet). The results show that the melody generated by the proposed model is closer to real music.

Deep learning application detecting SARS-CoV-2 key enzymes inhibitors

The fast spread of the COVID-19 over the world pressured scientists to find its cures. Especially, with the disastrous results, it engendered from human life losses to long-term impacts on infected people's health and the huge financial losses. In addition to the massive efforts made by researchers and medicals on finding safe, smart, fast, and efficient methods to accurately make an early diagnosis of the COVID-19. Some researchers focused on finding drugs to treat the disease and its symptoms, others worked on creating effective vaccines, while several concentrated on finding inhibitors for the key enzymes of the virus, to reduce its spreading and reproduction inside the human body. These enzymes' inhibitors are usually found in aliments, plants, fungi, or even in some drugs. Since these inhibitors slow and halt the replication of the virus in the human body, they can help fight it at an early stage saving the patient from death risk. Moreover, if the human body's immune system gets rid of the virus at the early stage it can be spared from the disastrous sequels it may leave inside the patient's body. Our research aims to find aliments and plants that are rich in these inhibitors. In this paper, we developed a deep learning application that is trained with various aliments, plants, and drugs to detect if a component contains SARS-CoV-2 key inhibitor(s) intending to help them find more sources containing these inhibitors. The application is trained to identify various sources rich in thirteen coronavirus-2 key inhibitors. The sources are currently just aliments, plants, and seeds and the identification is done by their names.

Label-free neural networks-based inverse lithography technology

Neural network-based inverse lithography technology (NNILT) has been used to improve the computational efficiency of large-scale mask optimization for advanced photolithography. NNILT is now mostly based on labels, and its performance is affected by the quality of labels. It is difficult for NNILT to achieve high performance and extrapolation ability for mask optimization without using labels. Here, we propose a label-free NNILT (LF-NNILT), which is implemented completely without labels and greatly improves the printability of the target layouts and the manufacturability of the synthesized masks compared to the traditional ILT. More importantly, the optimization speed of LF-NNILT is two orders of magnitude faster than the traditional ILT. Furthermore, LF-NNILT is simpler to implement and can achieve better solvers to support the development of advanced lithography.

Dendrobium candidum quality detection in both food and medicine agricultural product: Policy, status, and prospective

Dendrobium candidum (DC) is an agricultural product for both food and medicine. It has a variety of beneficial effects on the human body with antioxidant, anti-inflammatory, antitumor, enhancing immune function, and other pharmacological activities. Due to less natural distribution, harsh growth conditions, slow growth, low reproduction rate, and excessive logging, wild DC has been seriously damaged and listed as an endangered herbal medicine variety in China. At present, the quality of DC was uneven in the market, so it is very necessary to detect its quality. This article summarized the methods of DC quality detection with traditional and rapid nondestructive, and it also expounded the correlation between DC quality factor and endophytes, which provides a theoretical basis for a variety of rapid detection methods in macromolecules. At last, this article put forward a variety of rapid nondestructive detection methods based on the emission spectrum. In view of the complexity of molecular structure, the quality correlation established by spectral analysis was greatly affected by varieties and environment. We discussed the possibility of DC quality detection based on the molecular dynamic calculation and simulation mechanism. Also, a multimodal fusion method was proposed to detect the quality. The literature review suggests that it is very necessary to understand the structure performance relationship, kinetic properties, and reaction characteristics of chemical substances at the molecular level by means of molecular chemical calculation and simulation, to detect a certain substance more accurately. At the same time, several modes are combined to form complementarity, eliminate ambiguity, and uncertainty and fuse the information of multiple modes to obtain more accurate judgment results.

An experimental comparison of the widely used pre‐trained deep neural networks for image classification tasks towards revealing the promise of transfer‐learning

The easiest way to propose a solution based on deep neural networks is using the pre‐trained models through the transfer‐learning technique. Deep learning platforms provide various pre‐trained deep neural networks that can be easily applied for image classification tasks. So, “Which pre‐trained model provides the best performance for image classification tasks?” is a question that instinctively comes to mind and should be shed light on by the research community. To this end, we propose an experimental comparison of the six popular pre‐trained deep neural networks, namely, (i) VGG19, (ii) ResNet50, (iii) DenseNet201, (iv) MobileNetV2, (v) InceptionV3, and (vi) Xception by employing them through the transfer‐learning technique. Then, the proposed benchmark models were both trained and evaluated under the same configurations on two gold‐standard datasets, namely, (i) CIFAR‐10 and (ii) Stanford Dogs to benchmark them. Three evaluation metrics were employed to measure performance differences between the employed pre‐trained models as follows: (i) Accuracy, (ii) training duration, and (iii) inference time. The key findings that were obtained through the conducted a wide variety of experiments were discussed.

Mass data processing and multidimensional database management based on deep learning

With the rapid development of the Internet of Things, the requirements for massive data processing technology are getting higher and higher. Traditional computer data processing capabilities can no longer deliver fast, simple, and efficient data analysis and processing for today’s massive data processing due to the real-time, massive, polymorphic, and heterogeneous characteristics of Internet of Things data. Mass heterogeneous data of different types of subsystems in the Internet of Things need to be processed and stored uniformly, so the mass data processing method is required to be able to integrate multiple different networks, multiple data sources, and heterogeneous mass data and be able to perform processing on these data. Therefore, this article proposes massive data processing and multidimensional database management based on deep learning to meet the needs of contemporary society for massive data processing. This article has deeply studied the basic technical methods of massive data processing, including MapReduce technology, parallel data technology, database technology based on distributed memory databases, and distributed real-time database technology based on cloud computing technology, and constructed a massive data fusion algorithm based on deep learning. The model and the multidimensional online analytical processing model of the multidimensional database based on deep learning analyze the performance, scalability, load balancing, data query, and other aspects of the multidimensional database based on deep learning. It is concluded that the accuracy of multidimensional database query data is as high as 100%, and the accuracy of the average data query time is only 0.0053 s, which is much lower than the general database query time.

Artificial Intelligence Meets Whole Slide Images: Deep Learning Model Shapes an Immune-Hot Tumor and Guides Precision Therapy in Bladder Cancer

Background

To construct and validate a deep learning cluster from whole slide images (WSI) for depicting the immunophenotypes and functional heterogeneity of the tumor microenvironment (TME) in patients with bladder cancer (BLCA) and to explore an artificial intelligence (AI) score to explore the underlying biological pathways in the developed WSI cluster.

Methods

In this study, the WSI cluster was constructed based on a deep learning procedure. Further rerecognition of TME features in pathological images was applied based on a neural network. Then, we integrated the TCGA cohort and several external testing cohorts to explore and validate this novel WSI cluster and a corresponding quantitative indicator, the AI score. Finally, correlations between the AI cluster (AI score) and classical BLCA molecular subtypes, immunophenotypes, functional heterogeneity, and potential therapeutic method in BLCA were assessed.

Results

The WSI cluster was identified associated with clinical survival (P < 0.001) and was proved as an independent predictor (P = 0.031), which could also predict the immunology and the clinical significance of BLCA. Rerecognition of pathological images established a robust 3-year survival prediction model (with an average classification accuracy of 86%, AUC of 0.95) for BLCA patients combining TME features and clinical features. In addition, an AI score was constructed to quantify the underlying logic of the WSI cluster (AUC = 0.838). Finally, we hypothesized that high AI score shapes an immune-hot TME in BLCA. Thus, treatment options including immune checkpoint blockade (ICB), chemotherapy, and ERBB therapy can be used for the treatment of BLCA patients in WSI cluster1 (high AI score subtype).

Conclusions

In general, we showed that deep learning can predict prognosis and may aid in the precision medicine for BLCA directly from H&E histology, which is more economical and efficient.

Analysis of physical education based on deep learning on college students’ mental health and social adaptability

With the development of learning abroad, deep learning is used in research fields. On the basis of deep learning, this article studies physical education. First, this article analyzes and explains the related concepts and current situation of physical education, and explains the measurement and definition of the mental health. Then, the function analysis algorithm of deep learning is explained and analyzed, in which the algorithm of the convolution neural network of deep learning is mainly described. Finally, through experimental analysis, it shows that the research performance of deep learning in the physical education on college students’ mental health is relatively high. At the same time, through investigation and analysis, it is proposed that physical education in deep learning can improve mental health and social adaptability relatively high. And the content of physical education should focus on increasing physical psychological education and physical practice education, which can improve college students’ mental health and social adaptability compared with other teaching contents. Therefore, when introducing deep learning, universities should strengthen the physical education of college students.

Two Hours Ahead Prediction of the TEC over China Using a Deep Learning Method

In this paper, we forecast the regional total electron content (TEC) over China (0–60° N, 70–140° E) two hours in advance using a deep learning method called pix2pixhd that is based on Generative Adversarial Networks (GAN). We use the International GNSS Service (IGS) TEC maps over China during the 2003–2018 period for training and divide the data into three parts: a training set (2003–2013), a test set (2014–2017), and a validation set (2018). We evaluate the prediction effect of our model using Root Mean Square Error and correlation coefficient and compare our model with IRI-2016. The result demonstrates that our model shows a good performance for TEC prediction in China. Under different geomagnetic and solar activity conditions, the performance of our model is always better than IRI-2016. Analyzing the average difference map between the output of our model and the target IGS TEC map (+2 h), our model behaves well in China including the low-latitude region. In addition, our model behaves better during quiet time and high solar activity years. The successful application of pix2pixhd in forecasting the regional TEC maps over China demonstrates that deep learning methods can solve many geoscience problems, especially for ionospheric parameter forecasting.

Water quality prediction model using Gaussian process regression based on deep learning for carbon neutrality in papermaking wastewater treatment system

Wastewater recycling is the measure with enormous potentiality to achieve carbon neutrality in wastewater treatment plants. High-precision online monitoring can improve the stability of wastewater treatment system and help wastewater recycling. A new water quality prediction CSWLSTM-GPR model, which fused the spatial feature of convolutional neural network (CNN), the temporal feature of sharing-weight long short-term memory (SWLSTM) and the probabilistic reliability of Gaussian process regression (GPR), was applied for monitoring papermaking wastewater treatment system with high-precision point prediction and interval prediction. Compared with SWLSTM-GPR and CLSTM-GPR, RMSE of CSWLSTM-GPR reduced by more than 48.9% on effluent chemical oxygen demand (COD_eff), MAE reduced by more than 49.3%, R² increased by more than 25.14%, R increased by more than 7.07%. And for the effluent suspended solids (SS_eff), CSWLSTM-GPR had better predictive results than SWLSTM-GPR and CSWLSTM-GPR. Compared with SWLSTM-GPR, RMSE, MAE, R, R² of CSWLSTM-GPR on effluent suspended solids (SS_eff) were improved by 4.8%, 6.1%, 29.01% and 31.15%, respectively. Simulation results showed convincing comprehensive forecasting ability were obtained and the true values frequently stayed within the water quality range obtained by CSWLSTM-GPR model, which provided important insights for online monitoring, wastewater recycling and carbon neutrality of papermaking industry.

Identification of geographical origin and different parts of Wolfiporia cocos from Yunnan in China using PLS-DA and ResNet based on FT-NIR

INTRODUCTION

Wolfiporia cocos, as a kind of medicine food homologous fungus, is well-known and widely used in the world. Therefore, quality and safety have received worldwide attention, and there is a trend to identify the geographic origin of herbs with artificial intelligence technology.

OBJECTIVE

This research aimed to identify the geographical traceability for different parts of W. cocos.

METHODS

The exploratory analysis is executed by two multivariate statistical analysis methods. The two-dimensional correlation spectroscopy (2DCOS) images combined with residual convolutional neural network (ResNet) and partial least square discriminant analysis (PLS-DA) models were established to identify the different parts and regions of W. cocos. We compared and analysed 2DCOS images with different fingerprint bands including full band, 8900-6850 cm^-1 , 6300-5150 cm^-1 and 4450-4050 cm^-1 of original spectra and the second-order derivative (SD) spectra preprocessed.

RESULTS

From all results: the exploratory analysis results showed that t-distributed stochastic neighbour embedding was better than principal component analysis. The synchronous SD 2DCOS is more suitable for the identification and analysis of complex mixed systems for the small-band for Poria and Poriae cutis. Both models of PLS-DA and ResNet could successfully identify the geographical traceability of different parts based on different bands. The 10% external verification set of the ResNet model based on synchronous 2DCOS can be accurately identified.

CONCLUSION

Therefore, the methods could be applied for the identification of geographical origins of this fungus, which may provide technical support for quality evaluation.

An Integrated Change Detection Method Based on Spectral Unmixing and the CNN for Hyperspectral Imagery

Hyperspectral remote sensing image (HSI) include rich spectral information that can be very beneficial for change detection (CD) technology. Due to the existence of many mixed pixels, pixel-wise approaches can lead to considerable errors in the resulting CD map. The spectral unmixing (SU) method is a potential solution to this problem, as it decomposes mixed pixels into a set of fractions of land cover. Subsequently, the CD map is created by comparing the abundance images. However, based only on the abundance images created through the SU method, they are unable to effectively provide detailed change information. Meanwhile, the features of change information cannot be sufficiently extracted by the traditional sub-pixel CD framework, which leads to a poor CD result. To address these problems, this paper presents an integrated CD method based on multi-endmember spectral unmixing, joint matrix and CNN (MSUJMC) for HSI. Three main steps are considered to accomplish this task. First, considering the endmember spectral variability, more reliable endmember abundance information is obtained by multi-endmember spectral unmixing (MSU). Second, the original image features are incorporated with the abundance images using a joint matrix (JM) algorithm to provide more temporal and spatial land cover change information characteristics. Third, to efficiently extract the change features and to better handle the fused multi-source information, the convolutional neural network (CNN) is introduced to realize a high-accuracy CD result. The proposed method has been verified on simulated and real multitemporal HSI datasets, which provide multiple changes. Experimental results verify the effectiveness of the proposed approach.

Artificial intelligence in cancer target identification and drug discovery

Artificial intelligence is an advanced method to identify novel anticancer targets and discover novel drugs from biology networks because the networks can effectively preserve and quantify the interaction between components of cell systems underlying human diseases such as cancer. Here, we review and discuss how to employ artificial intelligence approaches to identify novel anticancer targets and discover drugs. First, we describe the scope of artificial intelligence biology analysis for novel anticancer target investigations. Second, we review and discuss the basic principles and theory of commonly used network-based and machine learning-based artificial intelligence algorithms. Finally, we showcase the applications of artificial intelligence approaches in cancer target identification and drug discovery. Taken together, the artificial intelligence models have provided us with a quantitative framework to study the relationship between network characteristics and cancer, thereby leading to the identification of potential anticancer targets and the discovery of novel drug candidates.

Big Data Fusion Method Based on Internet of Things Collection

With the development of information collection technology, the data that people need to deal with are also increasing, which brings problems such as isomerization of data types, poor data quality, and fast data generation speed. At present, as an important method of data fusion technology, data fusion method based on deep learning has become an effective way of data fusion under the background of big data, which has important research significance. There is a problem with heterogeneous data types between time series data and text data, and it is difficult to fuse them effectively by traditional data fusion methods. In order to make full use of the information contained in text data and improve the accuracy of time series prediction, this paper proposes a data fusion model based on FC-SAE. In this model, GloVe and CNN are used to extract the features of text data, FC neural network is used to extract the potential features of time series data, and then, the SEA model is used to fuse the data, which fully discovers the relationship between data and greatly improves the prediction accuracy.

Current research and future prospects of IVOCT imaging-based detection of the vascular lumen and vulnerable plaque

Intravascular optical coherence tomography (IVOCT) is an imaging method that has developed rapidly in recent years and is useful in coronary atherosclerosis diagnosis. It is widely used in the assessment of vulnerable plaque. This review summarizes the main research methods used in recent years for blood vessel lumen boundary detection and segmentation and vulnerable plaque segmentation and classification. This article aims to comprehensively and systematically introduce the research progress on internal tissues of blood vessels based on IVOCT images. The characteristics and advantages of various methods have been summarized to provide theoretical ideas and methods for the reference of relevant researchers and scholars.

Knowledge Management in Relationship Among Abusive Management, Self-Efficacy, and Corporate Performance Under Artificial Intelligence

The purpose is to explore the application potential of HCI (Human-Computer Interaction) technology under AI (Artificial Intelligence) in enterprise performance evaluation and the influence of abusive management and self-efficacy on enterprise performance. Guided by psychological theory, employees from a listed real estate enterprise are selected, and the research themes of abusive management, self-efficacy, and employee performance are assumed. Afterward, the employee job satisfaction and performance evaluation model and system interface based on deep learning BPNN (BackPropagation Neural Network), SVM (Support Vector Machine) regression, and HCI are innovatively proposed. The results show that the HCI interface can be accessed accurately according to the employee's verbal instructions. BPNN model has reached the best performance at the iteration of 70times, and all indexes have reached the expected employee satisfaction.

UAV Imagery for Automatic Multi-Element Recognition and Detection of Road Traffic Elements

Road traffic elements comprise an important part of roads and represent the main content involved in the construction of a basic traffic geographic information database, which is particularly important for the development of basic traffic geographic information. However, the following problems still exist for the extraction of traffic elements: insufficient data, complex scenarios, small targets, and incomplete element information. Therefore, a set of road traffic multielement remote sensing image datasets obtained by unmanned aerial vehicles (UAVs) is produced, and an improved YOLOv4 network algorithm combined with an attention mechanism is proposed to automatically recognize and detect multiple elements of road traffic in UAV imagery. First, the scale range of different objects in the datasets is counted, and then the size of the candidate box is obtained by the k-means clustering method. Second, mosaic data augmentation technology is used to increase the number of trained road traffic multielement datasets. Then, by integrating the efficient channel attention (ECA) mechanism into the two effective feature layers extracted from the YOLOv4 backbone network and the upsampling results, the network focuses on the feature information and then trains the datasets. At the same time, the complete intersection over union (CIoU) loss function is used to consider the geometric relationship between the object and the test object, to solve the overlapping problem of the juxtaposed dense test element anchor boxes, and to reduce the rate of missed detection. Finally, the mean average precision (mAP) is calculated to evaluate the experimental effect. The experimental results show that the mAP value of the proposed method is 90.45%, which is 15.80% better than the average accuracy of the original YOLOv4 network. The average detection accuracy of zebra crossings, bus stations, and roadside parking spaces is improved by 12.52%, 22.82%, and 12.09%, respectively. The comparison experiments and ablation experiments proved that the proposed method can realize the automatic recognition and detection of multiple elements of road traffic, and provide a new solution for constructing a basic traffic geographic information database.

Online reviews sentiment analysis and product feature improvement with deep learning

The text mining of online reviews is currently a popular research direction of e-commerce and is considered the next blue ocean. Online reviews can dig out consumer preferences and provide theoretical guidance for the improvement of product features. However, current research mostly focuses on sentiment analysis methods and rarely involves feature extraction and large-scale data recognition. This paper uses word segmentation technology to create a new feature extraction method. With long-short term memory(LSTM) neural network and latent dirichlet allocation(LDA) topic model, we proposes a product feature improvement model (Consumer online reviews-Extract short text-Sentiment analysis-Cluster feature, CESC). The model can derive the product features and attitudes that consumers prefer based on consumer online reviews, and use it to improve product features. According to the experimental results of three electronic products sold on the e-commerce platform, the model can effectively dig out consumer preferences for online reviews. Enterprises can improve the quality of products and services, better meet the needs of consumers, promote consumers’ consumption, and achieve the enterprises’ goals and values.

The Effects of Rso2 and PI Monitoring Images on the Treatment of Premature Infants Based on Deep Learning

In recent years, due to the combined effects of individual behavior, psychological factors, environmental exposure, medical conditions, biological factors, etc., the incidence of preterm birth has gradually increased, so the incidence of various complications of preterm infants has also become higher and higher. This article is aimed at studying the therapeutic effects of preterm infants and proposing the application of rSO2 and PI image monitoring based on deep learning to the treatment of preterm infants. This article introduces deep learning, blood perfusion index, preterm infants, and other related content in detail and conducts experiments on the treatment of rSO2 and PI monitoring images based on deep learning in preterm infants. The experimental results show that the rSO2 and PI monitoring images based on deep learning can provide great help for the treatment of preterm infants and greatly improve the treatment efficiency of preterm infants by at least 15%.

Inferring Metabolite-Disease Association Using Graph Convolutional Networks

As is well known, biological experiments are time-consuming and laborious, so there is absolutely no doubt that developing an effective computational model will help solve these problems. Most of computational models rely on the biological similarity and network-based methods that cannot consider the topological structures of metabolite-disease association graphs. We proposed a novel method based on graph convolutional networks to infer potential metabolite-disease association, named MDAGCN. We first calculated three kinds of metabolite similarities and three kinds of disease similarities. The final similarity of disease and metabolite will be obtained by integrating three kinds' similarities of each and filtering out the noise similarity values. Then metabolite similarity network, disease similarity network and known metabolite-disease association network were used to construct a heterogenous network. Finally, heterogeneous network with rich information is fed into the graph convolutional networks to obtain new features of a node through aggregation of node information so as to infer the potential associations between metabolites and diseases. Experimental results show that MDAGCN achieves more reliable results in cross validation and case studies when compared with other existing methods.

Design of Motion Detection Device in Sports Based on Deep Learning of Internet of Things

With the improvement of people's income levels in recent years, people have gradually begun to pay more attention to health, and the number of exercise and fitness people has increased year by year. People are gradually willing to pay for sports and fitness, increase sports consumption, and promote the development of the sports and fitness industry. This article aims to study the deep learning based on the Internet of things to make people aware of the importance of sports. Not loving sports is a major problem that contemporary people need to overcome. This article proposes how to design a motion detection device in sports based on deep learning of the Internet of things. Based on the calculation of the economic volume of the deep learning of the Internet of things and the questionnaire survey method, it can be seen that, in today's globalization, although everyone knows the importance of sports, they are unwilling to practice it and would rather spend more time on the Internet. The experimental results of this article show that more than 50% of college students are very interested in sports and fitness, but the actual use is less than 30%, which is not optimistic. In social surveys, this number will be even lower, with only 14% of people interested in sports. Big data is like a "double-edged sword." It not only displays the user's exercise data in front of everyone through the built-in sensors of the mobile phone, but also manages their physical condition through these. How to use the strengths of sports applications at the same time properly disposing of private information is a part of the next development of sports applications that must be faced.

Experimental Research on the Antitumor Effect of Human Gastric Cancer Cells Transplanted in Nude Mice Based on Deep Learning Combined with Spleen-Invigorating Chinese Medicine

Gastric cancer is still the fifth most common malignant tumor in the world and has the fourth highest mortality rate in the world. Gastric cancer is difficult to treat because of its unobvious onset, low resection rate, and rapid deterioration. Therefore, humans have been working hard to combat gastric cancer. At present, the most commonly used treatment method is radiotherapy. However, this method will damage the normal tissues of the irradiated area while treating malignant tumor cells. It not only has side effects of damage to the patient's skin and mucous membranes but also needs high-rate radiotherapy and has high cost for chemotherapy. In order to solve these problems, it is necessary to find new treatment methods. This article proposes the use of Chinese medicine to invigorate the spleen to inhibit human gastric cancer cells. This article combines modern machine learning technology with traditional Chinese medicine and combines traditional Chinese medicine physiotherapy with Western medicine nude mouse transplantation experiments. The treatment of tumors in Chinese medicine is based on the theory of Chinese medicine and has different characteristics. Western medicine has the advantage of permanently injuring patients. The process of the experiment is to transplant human-derived gastric cancer cells into nude mice. After grouping treatments and obtaining comparative data, deep learning techniques are used to analyze the properties of Chinese medicines for strengthening the spleen and to compare the properties of Chinese medicines for strengthening the spleen. The experimental results showed that the tumor inhibition rate of mice using fluorouracil was 18%, the tumor inhibition rate of mice using low-dose Chinese medicine was 16%, and the tumor inhibition rate of mice using high-dose Chinese medicine reached 52%. 80 days after the experiment, the survival rate of mice using high-dose Chinese medicine is 100% higher than that of mice without treatment.

Automatic Arrangement of Sports Dance Movement Based on Deep Learning

Sports dance is a new form of sports that integrates sports, dance, music, and other elements. The core content of “dance” is an important carrier for athletes to display their body art. This article aims to study the automatic arrangement of sports dance based on deep learning. This article first introduces the development process of deep learning. As the latest research direction developed from artificial neural network technology in machine learning, deep learning has attracted widespread attention from the society. And then proposing a shallow regression model based on deep learning, a convolutional neural network based on deep learning, and an offline sorting regression model, given the general process of deep learning, then, based on the clustering algorithm, the deep learning was researched, and the sport dance movement arrangement was analyzed based on the deep learning. The experimental results of this article show that deep learning can effectively enhance the artistic ability of automatic choreography in sports dance and increase the accuracy of dance movements by 80%. At the same time, on the basis of deep learning, the practical ability is strengthened on the basis of consolidating theory, to further improve one's own business ability and educational technology level, actively absorb advanced teaching methods, and earnestly delve into reasonable teaching methods. It is also used in curriculum training practice to actively gain insight into new development trends in educational methods and skills, to enhance the artistic creativity of students' arrangements.

Feature selection of infrared spectra analysis with convolutional neural network

Data-driven deep learning analysis, especially for convolution neural network (CNN), has been developed and successfully applied in many domains. CNN is regarded as a black box, and the main drawback is the lack of interpretation. In this study, an interpretable CNN model was presented for infrared data analysis. An ascending stepwise linear regression (ASLR)-based approach was leveraged to extract the informative neurons in the flatten layer from the trained model. The characteristic of CNN network was employed to visualize the active variables according to the extracted neurons. Partial least squares (PLS) model was presented for comparison on the performance of extracted features and model interpretation. The CNN models yielded accuracies with extracted features of 93.27%, 97.50% and 96.65% for Tablet, meat, and juice datasets on the test set, while the PLS-DA models obtained accuracies with latent variables (LVs) of 95.19%, 95.50% and 98.17%. Both the CNN and PLS models demonstrated the stable patterns on active variables. The repeatability of CNN model and proposed strategies were verified by conducting the Monte-Carlo cross-validation.

Deep Learning and Improved HMM Training Algorithm and Its Analysis in Facial Expression Recognition of Sports Athletes

Facial expressions are an auxiliary embodiment of information conveyed in the communication between people. Facial expressions can not only convey the semantic information that people want to express but also convey the emotional state of the speaker at the same time. But for sports athletes in training and competitions, it is usually not convenient to communicate directly. This paper is based on deep learning and an improved HMM training algorithm to study the facial expression recognition of sports athletes. It proposes the construction of deep learning of multilayer neural network, and the rank algorithm is introduced to carry out face recognition experiments with traditional HMM and class-specific HMM methods. The experimental results show that, with the increase of rank value, the class-specific recognition rate is up to 90%, the detection rate is 98% and the time-consuming is 2.5 min, which is better than HMM overall.

Data Reconstruction of Wireless Sensor Network and Zonal Demand Control in a Large-Scale Indoor Space Considering Thermal Coupling

An indoor high and open space is characterized by high mobility of people and uneven temperature distribution, so the conventional design and operation of air conditioning systems makes it difficult to regulate the air conditioning system precisely and efficiently. Thus, a Wireless Sensor Network was constructed in an indoor space located in Hong Kong to monitor the indoor environmental parameters of the space and improve the temperature control effectively. To ensure the continuity of the measurement data, three algorithms for reconstructing temperature, relative humidity and carbon dioxide data were implemented and compared. The results demonstrate the accuracy of support vector regression model and multiple linear regression model is higher than Back Propagation neural network model for reconstructing temperature data. Multiple linear regression is the most convenient from the perspective of program complexity, computing speed and difficulty in obtaining input conditions. Based on the data we collected, the traditional single-input-single-output control, zonal temperature control and the proposed zonal demand control methods were modeled on a Transient System Simulation Program (TRNSYS) control platform, the thermal coupling between the subzones without physical partition was taken into account, and the mass transfer between the virtual boundaries was calculated by an external CONTAM program. The simulation results showed the proposed zonal demand control can alleviate the over-cooling or over-heating phenomenon in conventional temperature control, thermal comfort and energy reduction is enhanced as well.

A comparative study of auto-contouring softwares in delineation of organs at risk in lung cancer and rectal cancer

Radiotherapy requires the target area and the organs at risk to be contoured on the CT image of the patient. During the process of organs-at-Risk (OAR) of the chest and abdomen, the doctor needs to contour at each CT image. The delineations of large and varied shapes are time-consuming and laborious. This study aims to evaluate the results of two automatic contouring softwares on OARs definition of CT images of lung cancer and rectal cancer patients. The CT images of 15 patients with rectal cancer and 15 patients with lung cancer were selected separately, and the organs at risk were manually contoured by experienced physicians as reference structures. And then the same datasets were automatically contoured based on AiContour (version 3.1.8.0, Manufactured by Linking MED, Beijing, China) and Raystation (version 4.7.5.4, Manufactured by Raysearch, Stockholm, Sweden) respectively. Deep learning auto-segmentations and Atlas were respectively performed with AiContour and Raystation. Overlap index (OI), Dice similarity index (DSC) and Volume difference (Dv) were evaluated based on the auto-contours, and independent-sample t-test analysis is applied to the results. The results of deep learning auto-segmentations on OI and DSC were better than that of Atlas with statistical difference. There was no significant difference in Dv between the results of two software. With deep learning auto-segmentations, auto-contouring results of most organs in the chest and abdomen are good, and with slight modification, it can meet the clinical requirements for planning. With Atlas, auto-contouring results in most OAR is not as good as deep learning auto-segmentations, and only the auto-contouring results of some organs can be used clinically after modification.

The Architecture of Mass Customization-Social Internet of Things System: Current Research Profile

In the era of big data, mass customization (MC) systems are faced with the complexities associated with information explosion and management control. Thus, it has become necessary to integrate the mass customization system and Social Internet of Things, in order to effectively connecting customers with enterprises. We should not only allow customers to participate in MC production throughout the whole process, but also allow enterprises to control all links throughout the whole information system. To gain a better understanding, this paper first describes the architecture of the proposed system from organizational and technological perspectives. Then, based on the nature of the Social Internet of Things, the main technological application of the mass customization–Social Internet of Things (MC–SIOT) system is introduced in detail. On this basis, the key problems faced by the mass customization–Social Internet of Things system are listed. Our findings are as follows: (1) MC–SIOT can realize convenient information queries and clearly understand the user’s intentions; (2) the system can predict the changing relationships among different technical fields and help enterprise R&D personnel to find technical knowledge; and (3) it can interconnect deep learning technology and digital twin technology to better maintain the operational state of the system. However, there exist some challenges relating to data management, knowledge discovery, and human–computer interaction, such as data quality management, few data samples, a lack of dynamic learning, labor consumption, and task scheduling. Therefore, we put forward possible improvements to be assessed, as well as privacy issues and emotional interactions to be further discussed, in future research. Finally, we illustrate the behavior and evolutionary mechanism of this system, both qualitatively and quantitatively. This provides some idea of how to address the current issues pertaining to mass customization systems.

EMCBOW-GPCR: A method for identifying G-protein coupled receptors based on word embedding and wordbooks

•

An computational method was developed to identify G-protein coupled receptors.

•

Three word-embedding models and a bag-of-words model are used to extract original features.

•

A high accuracy was achieved by using fusion information.

•

A powerful tool was established.

G Protein-Coupled Receptors (GPCRs) are one of the largest membrane protein receptor family in human, which are also important targets for many drugs. Thence, it’s of great significance to judge whether a protein is a GPCR or not. However, identifying GPCRs by experimental methods is very expensive and time-consuming. As more and more GPCR primary sequences are accumulated, it’s feasible to develop a computational model to predict GPCRs precisely and quickly. In this paper, a novel method called EMCBOW-GPCR has been proposed to improve the accuracy of identifying GPCRs based on natural language processing (NLP). For representing GPCRs, three word-embedding models and a bag-of-words model are used to extract original features. Then, the original features are thrown into a Deep-learning algorithm to extract features further and reduce the dimension. Finally, the obtained features are fed into Extreme Gradient Boosting. As shown with the results comparison, the overall prediction metrics of EMCBOW-GPCR are higher than the state of the arts. In order to be convenient for more researchers to use EMCBOW-GPCR, the method and source code have been opened in github, which are available at https://github.com/454170054/EMCBOW-GPCR, and a user-friendly web-server for EMCBOW-GPCR has been established at http://www.jci-bioinfo.cn/emcbowgpcr.

Classification of Geometric Forms in Mosaics Using Deep Neural Network

The paper addresses an image processing problem in the field of fine arts. In particular, a deep learning-based technique to classify geometric forms of artworks, such as paintings and mosaics, is presented. We proposed and tested a convolutional neural network (CNN)-based framework that autonomously quantifies the feature map and classifies it. Convolution, pooling and dense layers are three distinct categories of levels that generate attributes from the dataset images by introducing certain specified filters. As a case study, a Roman mosaic is considered, which is digitally reconstructed by close-range photogrammetry based on standard photos. During the digital transformation from a 2D perspective view of the mosaic into an orthophoto, each photo is rectified (i.e., it is an orthogonal projection of the real photo on the plane of the mosaic). Image samples of the geometric forms, e.g., triangles, squares, circles, octagons and leaves, even if they are partially deformed, were extracted from both the original and the rectified photos and originated the dataset for testing the CNN-based approach. The proposed method has proved to be robust enough to analyze the mosaic geometric forms, with an accuracy higher than 97%. Furthermore, the performance of the proposed method was compared with standard deep learning frameworks. Due to the promising results, this method can be applied to many other pattern identification problems related to artworks.

Soccer-Assisted Training Robot Based on Image Recognition Omnidirectional Movement

With the continuous emergence and innovation of computer technology, mobile robots are a relatively hot topic in the field of artificial intelligence. It is an important research area of more and more scholars. The core of mobile robots is to be able to realize real-time perception of the surrounding environment and self-positioning and to conduct self-navigation through this information. It is the key to the robot’s autonomous movement and has strategic research significance. Among them, the goal recognition ability of the soccer robot vision system is the basis of robot path planning, motion control, and collaborative task completion. The main recognition task in the vision system is the omnidirectional vision system. Therefore, how to improve the accuracy of target recognition and the light adaptive ability of the robot omnidirectional vision system is the key issue of this paper. Completed the system construction and program debugging of the omnidirectional mobile robot platform, and tested its omnidirectional mobile function, positioning and map construction capabilities in the corridor and indoor environment, global navigation function in the indoor environment, and local obstacle avoidance function. How to use the local visual information of the robot more perfectly to obtain more available information, so that the “eyes” of the robot can be greatly improved by relying on image recognition technology, so that the robot can obtain more accurate environmental information by itself has always been domestic and foreign one of the goals of the joint efforts of scholars. Research shows that the standard error of the experimental group’s shooting and dribbling test scores before and the experimental group’s shooting and dribbling test results after the standard error level is 0.004, which is less than 0.05, which proves the use of soccer-assisted robot-assisted training. On the one hand, we tested the positioning and navigation functions of the omnidirectional mobile robot, and on the other hand, we verified the feasibility of positioning and navigation algorithms and multisensor fusion algorithms.

Research on insulator defect detection algorithm of transmission line based on CenterNet

The reliability of the insulator has directly affected the stable operation of electric power system. The detection of defective insulators has always been an important issue in smart grid systems. However, the traditional transmission line detection method has low accuracy and poor real-time performance. We present an insulator defect detection method based on CenterNet. In order to improve detection efficiency, we simplified the backbone network. In addition, an attention mechanism is utilized to suppress useless information and improve the accuracy of network detection. In image preprocessing, the blurring of some detected images results in the samples being discarded, so we use super-resolution reconstruction algorithm to reconstruct the blurred images to enhance the dataset. The results show that the AP of the proposed method reaches 96.16% and the reasoning speed reaches 30FPS under the test condition of NVIDIA GTX 1080 test conditions. Compared with Faster R-CNN, YOLOV3, RetinaNet and FSAF, the detection accuracy of proposed method is greatly improved, which fully proves the effectiveness of the proposed method.

Discovering generative models from event logs: data-driven simulation vs deep learning

A generative model is a statistical model capable of generating new data instances from previously observed ones. In the context of business processes, a generative model creates new execution traces from a set of historical traces, also known as an event log. Two types of generative business process models have been developed in previous work: data-driven simulation models and deep learning models. Until now, these two approaches have evolved independently, and their relative performance has not been studied. This paper fills this gap by empirically comparing a data-driven simulation approach with multiple deep learning approaches for building generative business process models. The study sheds light on the relative strengths of these two approaches and raises the prospect of developing hybrid approaches that combine these strengths.

A robust identification method for nonferrous metal scraps based on deep learning and superpixel optimization

End-of-life vehicles (ELVs) provide a particularly potent source of supply for metals. Hence, the recycling and sorting techniques for ferrous and nonferrous metal scraps from ELVs significantly increase metal resource utilization. However, different kinds of nonferrous metal scraps, such as aluminium (Al) and copper (Cu), are not further automatically classified due to the lack of proper techniques. The purpose of this study is to propose an identification method for different nonferrous metal scraps, facilitate the further separation of nonferrous metal scraps, achieve better management of recycled metal resources and increase sustainability. A convolutional neural network (CNN) and SEEDS (superpixels extracted via energy-driven sampling) were adopted in this study. To build the classifier, 80 training images of randomly chosen Al and Cu scraps were taken, and some practical methods were proposed, including training patch generation with SEEDS, image data augmentation and automatic labelling methods for enormous training data. To obtain more accurate results, SEEDS was also used to optimize the coarse results obtained from the pretrained CNN model. Five indicators were adopted to evaluate the final identification results. Furthermore, 15 test samples concerning different classification environments were tested through the proposed model, and it performed well under all of the employed evaluation indexes, with an average precision of 0.98. The results demonstrate that the proposed model is robust for metal scrap identification, which can be expanded to a complex industrial environment, and it presents new possibilities for highly accurate automatic nonferrous metal scrap classification.

Trace Identification and Visualization of Multiple Benzimidazole Pesticide Residues on Toona sinensis Leaves Using Terahertz Imaging Combined with Deep Learning

Molecular spectroscopy has been widely used to identify pesticides. The main limitation of this approach is the difficulty of identifying pesticides with similar molecular structures. When these pesticide residues are in trace and mixed states in plants, it poses great challenges for practical identification. This study proposed a state-of-the-art method for the rapid identification of trace (10 mg·L⁻¹) and multiple similar benzimidazole pesticide residues on the surface of Toona sinensis leaves, mainly including benzoyl (BNL), carbendazim (BCM), thiabendazole (TBZ), and their mixtures. The new method combines high-throughput terahertz (THz) imaging technology with a deep learning framework. To further improve the model reliability beyond the THz fingerprint peaks (BNL: 0.70, 1.07, 2.20 THz; BCM: 1.16, 1.35, 2.32 THz; TBZ: 0.92, 1.24, 1.66, 1.95, 2.58 THz), we extracted the absorption spectra in frequencies of 0.2–2.2 THz from images as the input to the deep convolution neural network (DCNN). Compared with fuzzy Sammon clustering and four back-propagation neural network (BPNN) models (TrainCGB, TrainCGF, TrainCGP, and TrainRP), DCNN achieved the highest prediction accuracies of 100%, 94.51%, 96.26%, 94.64%, 98.81%, 94.90%, 96.17%, and 96.99% for the control check group, BNL, BCM, TBZ, BNL + BCM, BNL + TBZ, BCM + TBZ, and BNL + BCM + TBZ, respectively. Taking advantage of THz imaging and DCNN, the image visualization of pesticide distribution and residue types on leaves was realized simultaneously. The results demonstrated that THz imaging and deep learning can be potentially adopted for rapid-sensing detection of trace multi-residues on leaf surfaces, which is of great significance for agriculture and food safety.