Search

Coix seeds from 9 different origins were taken as the research object. An attempt was made to achieve geographical origin traceability of Coix seeds through the combination of excitation-emission matrix (EEM) fluorescence spectroscopy with improved random forest algorithm. The improvements to the random forest algorithm mainly include two aspects, firstly, principal component analysis (PCA) was adopted to reduce the dimension of EEM fluorescent data; secondly, a grid search method was used to identify the optimal number of principal components(PCs) to retain and the hyperparameters of the discriminant model during the PCA dimension reduction process. The results showed that an improved random forest model, incorporating standard deviation normalization and PCA dimension reduction modules, based on Coix seeds EEM fluorescence spectroscopy data, accurately predicted the geographical origin of Coix seed samples from 9 different areas. The optimal model was constructed by combining 100 decision trees with a maximum depth of 3 and a minimum sample size of 1 at the leaf node, using 16 principal components (PCs). This model achieved 100% prediction accuracy for both the validation and test sets, which consisted of a total of 108 samples, outperforming the PLS-DA model constructed by the partial least squares method (96% prediction accuracy). [ABSTRACT FROM AUTHOR]

The spatial patterns of star-rated rural homestays have significant implications for achieving high-quality rural tourism development. By employing methods such as nearest neighbor index, kernel density estimation, and decision tree regression, this research explores the spatial differentiation characteristics and causes of 330 star-rated rural homestays in the Wuling Mountain Area. The findings are as follows: 1) From the perspective of spatial structural features, star-rated rural homestays in the Wuling Mountain Area exhibit an overall "clustered in the east, scattered in the west, more in the north than in the south" spatial pattern. The distribution characteristics of four-star and five-star rural homestays are consistent with the overall pattern, while three-star rural homestays show an "eastern scarcity, western abundance, northern aggregation, and southern dispersion" spatial structure. 2) In terms of spatial typology, star-rated rural homestays in the Wuling Mountain Area display a cohesive spatial type. Specifically, five-star and four-star rural homestays exhibit a strongly cohesive spatial distribution, while three-star rural homestays show a weakly cohesive pattern. 3) Concerning spatial density features, the Wuling Mountain Area demonstrates a spatial density pattern where there are multiple and contiThe spatial patterns of star-rated rural homestays have significant implications for achieving high-quality rural tourism development. By employing methods such as nearest neighbor index, kernel density estimation, and decision tree regression, this research explores the spatial differentiation characteristics and causes of 330 star-rated rural homestays in the Wuling Mountain Area. The findings are as follows: 1) From the perspective of spatial structural features, star-rated rural homestays in the Wuling Mountain Area exhibit an overall "clustered in the east, scattered in the west, more in the north than in the south" spatial pattern. The distribution characteristics of four-star and five-star rural homestays are consistent with the overall pattern, while three-star rural homestays show an "eastern scarcity, western abundance, northern aggregation, and southern dispersion" spatial structure. 2) In terms of spatial typology, star-rated rural homestays in the Wuling Mountain Area display a cohesive spatial type. Specifically, five-star and four-star rural homestays exhibit a strongly cohesive spatial distribution, while three-star rural homestays show a weakly cohesive pattern. 3) Concerning spatial density features, the Wuling Mountain Area demonstrates a spatial density pattern where there are multiple and contiguous cores in the northern part and fewer and dispersed cores in the southern part. This pattern remains consistent for three-star and four-star rural homestays, while the five-star rural homestays exhibit a spatial density characteristic of "three scattered cores in the north and a single weak core in the south. " 4) The regression results of the decision tree model indicate that tourist attractions with a grade of 3A or higher (the root node), economic development level (branch node), and rural homestays policies (branch node) are important factors contributing to the spatial differentiation of star-rated rural homestays in the Wuling Mountain Area. The findings of this paper can provide reference for the layout of rural homestays in the Wuling Mountain Area and similar mountainous areas.guous cores in the northern part and fewer and dispersed cores in the southern part. This pattern remains consistent for three-star and four-star rural homestays, while the five-star rural homestays exhibit a spatial density characteristic of "three scattered cores in the north and a single weak core in the south. " 4) The regression results of the decision tree model indicate that tourist attractions with a grade of 3A or higher (the root node), economic development level (branch node), and rural homestays policies (branch node) are important factors contributing to the spatial differentiation of star-rated rural homestays in the Wuling Mountain Area. The findings of this paper can provide reference for the layout of rural homestays in the Wuling Mountain Area and similar mountainous areas. [ABSTRACT FROM AUTHOR]

Accurate monitoring of crop spatial distribution in topographically complex areas is of great significance for guiding agricultural management production and reasonable allocation of resources. However, scattered crop distribution and high spatial heterogeneity pose challenges for precise classification. The objective of this study was to explore a method for simultaneous precise identification of multiple crops with multi-temporal high-resolution satellite data under complex terrain conditions, thus further providing a theoretical and practical basis for crop identification in Chongqing city area and the complex low hills area in southwest China. The western Chongqing was selected as the study area. First, a multi-scale segmentation algorithm was adopted to construct homogeneous units from fields of the same feature type. This approach helped to achieve high accuracy classification by avoiding overly fragmented classification results based on pixel classification. Secondly, the multi-temporal Sentinel-2/MSI remote sensing images were used to explore the weathering patterns and characteristic parameters during the fertility period of major food crops. The spectral indices, such as NDVI, RVI and NDWI, were constructed to compare the differences in remote sensing spectral information of different crop types at various fertility periods based on ground sample points. Such information was then combined with the topographic features to determine the optimal combination for crop identification. Finally, an object-oriented decision tree logical classification rule set was established to extract major grain crop growing areas. The results showed that, (1) the multi-scale segmentation method was an effective approach that can make crop identification on a field basis. In addition, the vegetation edge segmentation was optimal at a segmentation scale of 30 and a compactness and shape factor of 0.5. (2) The target feature discrimination threshold was established by combining the spectral feature indices of NDWI in April, RVI in June, NDVI in August and NDVI in August with the topographic features such as altitude and slope. The overall accuracy of the classification reached 90.8%, being 85.7%, 83.3% and 80.7%, for paddy rice, maize, and sweet potato, respectively. This paper showed that the crop planting area identification method based on multiple spectral indices combined with topographic features can achieve high recognition accuracy and has practical significance. [ABSTRACT FROM AUTHOR]

Objectives: The key phenological information of oilseed rapeseed (Brassica napus L.) plays an important role in field management, viewing time prediction and yield estimation of the oilseed rape. It is also an important part of precision agriculture. Polarimetric synthetic aperture radar technology shows great potential in phenological phase identification with its all-weather monitoring capability and its sensitivity to the crop structural information. Methods: First, we identified the 5 phenological phases of the oilseed rape on the test area with 5 time series full-polarization Radarsat-2 data, which covers the whole growth period of the oilseed rape. 6 typical Stokes parameters are extracted and applied in the identification of oilseed rape phenological phases, the extracted Stokes parameters includ averaged intensity(g0), normalized average intensity(g0m), averaged degree of polarization(ρm), perimeter degree of zero orientation route(Pdor),inclination degree of zero aperture route(Idap), and arc asymmetry degree of zero aperture route(Aadap). Then, The phenological phases of oilseed rape is identified by the decision tree (DT) algorithm based on the comparative analysis of the dynamic response of the 6 special Stokes parameters to rape growth stages. Results and Conclusions: Among the extracted Stokes parameters applied in this study, except ρm and Aadap, other parameters show great sensitivity to the change of the oilseed rape phenological phases. The DT algorithm also perform well in the classification of the oilseed rape phenological phases. The classification results agree well with the field measured samples, and the overall classification accuracy is 87.4%, while the highest classifi⁃ cation accuracy of each phenological phase is 94.3%. [ABSTRACT FROM AUTHOR]

To meet the needs of real-time driving ability evaluation for human-machine co-driving decision problems for intelligent assisted driving systems， a method for real-time driving ability evaluation of drivers considering driving skill， driving state and driving style is proposed taking into account the unicity problem of existing driving evaluation researches. Based on the relative and continuous attributes of driving ability， firstly， an objective entropy-weighted relative evaluation model of driving skill is proposed based on the Gaussian kernel function， the relative evaluation model of driving state based on the time scale， and the soft classification model of driving style based on unsupervised decision classification tree. Secondly， a real-time driving ability evaluation mechanism and evaluation model with “punishment” and “affirmation” mechanisms are proposed to achieve real-time driving ability evaluation that meets the needs of human-machine shared decision control. Finally， the experimental comparison analysis shows that the proposed evaluation algorithm can meet the real-time， objective， and comprehensive requirements of human-machine co-driving decision control for drivers driving ability evaluation. [ABSTRACT FROM AUTHOR]

With the continuous rise of federated learning, gradient boosting decision tree（GBDT）,as a traditional machine learning method, is gradually applied to federated learning to achieve ideal classification results.Aiming at the problems of the existing horizontal federated learning model of GBDT,such as the accuracy is greatly influenced by non-IID dataset, information leakage and high communication cost, the paper proposed a federated gradient boosting decision tree for non-IID dataset, called nFL-GBDT.Firstly, the algorithm calculated similar samples among the participants by utilizing the LSH,and using weighted gradient constructed the first tree.Secondly, a reliable third party calculated the global leaf weight that only needed one round of communication.Finally, the experimental analysis shows that the algorithm can protect the privacy of the original data, and its communication cost is lower than that of simFL and FederBoost.At the same time, the experiment divides three groups of public data sets according to the imbalance ratio.The results show that the accuracy of this algorithm is improved by 3.53%,5.46% and 4.43% respectively compared with Individual, TFL and F-GBDT-G. [ABSTRACT FROM AUTHOR]

SMP Negeri 1 Soreang is one of the favorite public junior high schools in the Soreang area, Bandung Regency. To improve the quality of schools, it is done through the distribution of scholarships. Determination of scholarships is determined by using an assessment that has certain criteria. By utilizing classification techniques, one of which is the Decision Tree algorithm model, which is chosen to be able to assist in making decisions about who is eligible to get a scholarship so that it is right on target. Decision Tree Algorithm is a classification method to utilize data in order to find the relationship between the target variable and the input variable. The data contains student scores of 237 people. From the modeling process, the accuracy value is 92%. So it can be concluded that this algorithm can assist in providing selection decisions for junior high school (SMP) scholarship recipients. The Decision Tree algorithm was chosen because it has a high level of accuracy compared to other classification algorithms. In addition, the Decision Tree algorithm is used to make a decision from the information obtained. [ABSTRACT FROM AUTHOR]

As a critical part of desision tree (DT） growth, its nodes can be split to grow by either axis- or nonaxis-aligned way based on such splitting criteria as purity and misclassification error. However, these have nothing to do with the geometric structure of data, e. g. multicentric data or single-center data. In order to compensate for this, two splitting criteria are proposed by combining the between-class margin in the same inner node (BCM） and the between-node margin within the same class (BNM） respectively with the purity measure in weighting and the two-step method. Unlike traditional greedy growth of DT which only finds the current locally optimal splitting point, the proposed method first selects the top-k purity splitting nodes, then determines the optimal one by maximizing BCM and minimizing BNM. Since not only alleviating purity-based local optimality but also considering global structures of the data, our method greatly improves the division of descendant nodes and generalization of the formed trees, while enhancing the interpretability. In addition, two aforementioned splitting criteria can be combined to further boost the performance. The comparison results on 21 benchmark datasets show an improvement in predictive performance of new trees with reduction in complexity, while also are competitive with many other DTs using hybrid splitting criteria. [ABSTRACT FROM AUTHOR]

OBJECTIVE: To scientifically evaluate the economics of Longshengzhi capsules in the treatment of cerebral infarction (Qi deficiency and blood stasis syndrome) based on cost-effectiveness analysis, so as to provide reference for clinical drug use, academic promotion and decision-making of national pharmaceutical management departments. METHODS: From the perspective of health system, based on the evidence-based medical evidence of the safety and efficacy of Longshengzhi capsules in the treatment of cerebral infarction (Qi deficiency and blood stasis syndrome), the decision tree model of Longshengzhi capsules in the treatment of cerebral infarction was constructed by TreeAge Pro software. Economics of Longshengzhi capsules combined with conventional treatment in patients with cerebral infarction were explored. RESULTS: Combined with evidence-based medicine, drug instructions, drug nonclinical safety studies, and pharmacological and pharmacodynamic studies, it was shown that Longshengzhi capsules were safe in the treatment of patients with cerebral infarction. Meta-analysis of effectiveness of Longshengzhi capsules in the treatment of cerebral infarction showed that Longshengzhi capsules combined with conventional western medicine was superior to conventional western medicine alone in improving the National Institutes of Health Stroke Scale (NIHSS) score, quality of life score and serum homocysteine level in patients with cerebral infarction. Based on the above safety and efficacy evidence, the economic decision tree model simulation results showed that the incremental cost-effectiveness ratio of Longshengzhi capsules combined with conventional western medicine in improving the NIHSS score of patients with cerebral infarction was 238. 84 yuan / point, accounting for 0. 74% of per capita disposable income in 2020 in China ( 32 189 yuan), which had good economy. The result verified to be relatively stable. CONCLUSIONS: Longshengzhi capsules combined with conventional western medicine is more economical than conventional western medicine in improving the NIHSS score of patients with cerebral infarction (Qi deficiency and blood stasis syndrome). [ABSTRACT FROM AUTHOR]

BACKGROUND: The arch of the foot plays a supporting and cushioning role in people’s daily physical activities. The abnormal arch of the foot can cause movement disorders and lower limb pain. It is the premise of making the corresponding prevention and nursing corrective measures to accurately identify the pathological foot type. OBJECTIVE: The accuracy of foot recognition can be improved effectively by extracting common features of quadrupedal foot and designing foot interruption feature combined with gradient boosted decision tree classifier. METHODS: Totally 1 710 plantar pressure images of 45 people were collected, including high-arch, flat and normal foot. Arch index, footprint index, arch width, and ratio index of plantar pressure images were extracted respectively. At the same time, the foot interruption feature was designed, and the gradient boosted decision tree algorithm was used to recognize different foot types of plantar pressure images.RESULTS AND CONCLUSION: On the constructed data set of 1 710 plantar pressure images of 45 people, the average recognition accuracy of the proposed algorithm reached 96.43%, which was higher than the current commonly used methods based on arch index, footprint index, arch width, and ratio index. [ABSTRACT FROM AUTHOR]

OBJECTIVE: To evaluate the economical efficiency of Shexiang Baoxin pills, compound Danshen dropping pills and Naoxintong capsules in the treatment of stable angina pectoris of coronary heart disease. METHODS: Based on the perspective of health system, a decision tree model was constructed, and cost-utility analysis was used to conduct pharmacoeconomic evaluation on Shexiang Baoxin pills, compound Danshen dripping pills and Naoxintong capsules in the treatment of stable angina pectoris of coronary heart disease. RESULTS: Compared with the Shexiang Baoxin pills group, the incremental cost-utility ratio of the Naoxintong capsules group regimen was 3 501 593. 41, which was greater than 3 times the GDP per capita, so that the Shexiang Baoxin pills group regimen was more economical; compared with the compound Danshin dropping pills group, the Shexiang Baoxin pills group regimen was less costly and more effective; therefore, the Shexiang Baoxin pills group regimen was the most economical. Sensitivity analysis showed that the results were robust. CONCLUSIONS: Compared with compound Danshin dropping pills and Naoxintong capsules, the Shexiang Baoxin pills has more pharmacoeconomic advantages in the treatment of stable angina pectoris in coronary heart disease. [ABSTRACT FROM AUTHOR]

The deep decision tree migration learning boosting method ( DtrBoost) can effectively realize the migration learning from a single source domain to a target domain under supervision, but can not realize the unsupervised migration scenario under multiple source domains. To solve this problem, this paper proposed an unsupervised transfer learning boosting method for optimizing the weight under multi-source domain distribution. The main idea was to calculate the corresponding KL value according to the distribution of different source domains and target domains, selected an appropriate number of samples from different source domains to train the classifier and pseudo label the samples from the target domain. Finally, the algorithm assigned different learning weights according to the KL distance of each different source domain, and the labeled source domain samples integrated to the pseudo labeled target domain to obtain the final result. Comparative experiments show that the proposed algorithm achieves better classification accuracy and adaptive effect on different data sets. The average classification error rate decreases by 2. 4% and more than 6% on the best marketing data set. [ABSTRACT FROM AUTHOR]

Accurate identification of lettuce maturity is one of the most important steps during vegetable production. The better quality of lettuce can depend mainly on the picking time in the period of suitable maturity. Among them, the harvested lettuce texture is one of the important bases to determine lettuce maturity in this case. Fortunately, the mechanical characteristics can also be an optimal way to evaluate and detect the lettuce texture after harvesting. It is very necessary to clarify the relationship between the lettuce maturity and harvested texture. In this study, a novel classification was proposed to characterize the maturity of vegetables, according to the mechanical characteristics of hydroponic lettuce under the puncture test. Three categories were also classified by the immature, mature, and over-mature plants. Firstly, a systematic extraction was performed on the mechanical characteristics of the stem and mesophyll in the inner, middle, and outer leaves of immature, mature, and over mature plants in the time and frequency domain. The correlation analysis was made between the leaf mechanical characteristics and chlorophyll content, and the above-ground fresh weight. Some indicators were then calculated to accurately characterize the leaf maturity characteristics. A consistency analysis was also implemented between the mechanical characteristics and the changes of soluble sugar or soluble protein. Secondly, a dual-threshold depth traversal was designed to classify the individual characteristics of a single leaf and single region. The classification thresholds of immature and mature were obtained for the mature and over-mature mechanical characteristics. An optimal selection was performed on the blade type and region with the highest classification accuracy. All the mechanical features of the region were then trained using various machine learnings. The test results showed that the mechanical characteristics of the leaves were significantly correlated with the above-ground fresh quality of lettuce. There was a significantly low correlation with the Soil Plant Analysis Development (SPAD) value of the middle leaves, indicating the better consistence with the change law of soluble sugar and soluble protein. It infers that the mechanical characteristics of leaves can be expected to characterize the maturity of lettuce. The overall accuracy rate of the mechanical features in the single threshold classification was ranked in the descending order of: middle leaf > inner leaf > outer leaf. More importantly, the classification effect of each feature in the stem part was outstandingly better than that of the mesophyll part. Among them, the highest classification accuracy was 75.5% in the fracture force of the middle leaf stem. Specifically, the classification thresholds of fracture force were 0.98, and 1.38 N, respectively, for the immature and mature, while the mature and overripe plants. An optimal ensemble was achieved to better classify all the mechanical characteristics of the stem and mesophyll in the middle leaves. Correspondingly, the integration method was the Bagging, and the training learner was a decision tree. The accuracy rate was more than 94.3% for the three types of lettuce. A series of experiments were carried out to verify under the same conditions. Another batch of lettuce was planted in this case. The integrated model was also optimized using machine learning after training. The prediction accuracy rate was obtained by 91.7%, indicating the better validity of the improved model. Consequently, the mechanical characteristics can be expected to serve as a new tool for the rapid identification and accurate classification of vegetable maturity. An optimal harvest texture of mature lettuce was also determined to improve the harvest quality of lettuce. [ABSTRACT FROM AUTHOR]

Soil Organic Matter (SOM) is the carrier of soil organic carbon in the crop system. This component of soil solid matter is one of the core elements to evaluate soil fertility quality in agriculture and land management. An accurate and efficient acquisition of SOM content can greatly contribute to the quality grading of cultivated land. High-resolution remote sensing and Google Earth Engine (GEE) can serve as the computing platform for the efficient inversion of SOM. Much effort has been made on the SOM prediction model and the spatial distribution map. However, it is still lacking in the appropriate satellite data sources and prediction algorithms to accurately predict the SOM content in specific regions. In this study, the accurate SOC content was predicted in the cultivated land using GEE and machine learning. The Sentinel-2A MSI and the Landsat8 OLI data were collected in the Gaocheng District, Shijiazhuang City, Hebei Province, China. The main data sources were also combined with the Sentinel-1 SAR, ECMWF/ERA5 meteorological, and USGS/SRTMGL1_003 elevation data. The variable feature sets of the spectral band were constructed, including the vegetable index (Normalized Difference Vegetation Index (NDVI)；Red Index (RI)；Enhanced Vegetable Index (EVI)；Soil-Adjusted Total Vegetation Index (SATVI)；Brightness Index (BI)), radar feature (Sentinel-1 VV, and Sentinel-1 VH), terrain feature (slope, aspect, and elevation), and climate feature (annual precipitation, and average annual temperature). Six and five models were constructed using the Sentinel-2 and Landsat8 variable datasets, respectively. Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and Support Vector Machine (SVM) were utilized to predict the SOM on the GEE platform. The predictive performances of three machine learning methods were determined for a high-precision spatial distribution map for the SOM inversion. The accuracy of the prediction model was then evaluated using the determination coefficient (R²) and the root mean square error (RMSE). The results show that: 1) the R² and RMSE values of the model using the Sentinel-2A were better than those using the Landsat8. The Sentinel-2A model performed better than the Landsat8 model in the predicting SOM content. The best performance (R2 =0.759, RMSE=2.852 g/kg) was achieved in the omnivariate model of Sentinel-2A under the GBDT. 2) The maximum improvement of 9.752% was obtained in model A-1 with the red edge band, compared with model A-0. This difference was attributed to the inclusion of four red edge bands (B5, B6, B7, and B8A) in model A-1. The addition of red edge bands greatly improved the prediction accuracy of the model, particularly with the effective spectral information for the SOM inversion. 3) The red edge band, vegetable index, Sentinel-1A radar features, terrain factors, and climate variables greatly contributed to the prediction accuracy of SOM from the perspective of different variable feature combinations. 4) The GBDT was better applied to the SOM prediction in the study area. The resultant SOM map was used to accurately characterize the SOM spatial distribution. The test data was verified for high accuracy, each group of which was an excellent consistency in the image, indicating the reliable SOM inversion. Therefore, the Sentinel-2A MSI data presented outstanding advantages over the Landsat8 OLI, due to the higher spectral and spatial resolutions. The combination of GBDT, Sentinel-2A, and GEE can be an effective way to predict the SOM map. Each prediction factor can also provide valuable information for the prediction of SOM content. [ABSTRACT FROM AUTHOR]

OBJECTIVE: To explore the optimal drug treatment plan of azithromycin sequential therapy and azithromycin-minocycline therapy in the treatment of Mycoplasma pneumoniae pneumonia by using the pharmacoeconomic analysis methods, and obtain the best effect at the lowest cost. METHODS: Retrospective analysis was performed on 322 children with M. pneumoniae pneumonia over 8 years old admitted into pediatrics of the hospital from Aug. 2016 to Dec. 2018. According to the treatment regimen, all patients were divided into group A (azithromycin sequential therapy) with 160 cases and group B ( azithromycin-minocycline therapy) with 162 cases. Through decision tree and SPSS software, cost-effectiveness analysis of two groups was carried out. RESULTS: The improvement time of cough, fever, pulmonary signs and average length of stay in group A were significantly longer than those in group B, with statistically significant differences (P<0. 05). The total effective rates in group A and group B were respectively 89. 4% (143 / 160) and 92. 6% (150 / 162), with no statistically significant difference (P = 0. 529). The curing costs of one case with M. pneumoniae pneumonia in group A and group B were respectively 3 840 yuan and 3 570 yuan. No obvious adverse drug reactions were found in two groups. CONCLUSIONS: For children with M. pneumoniae pneumonia over 8 years old, there is no significant difference between azithromycin sequential therapy and azithromycin-minocycline therapy, and the cost of azithromycin-minocycline therapy is low. [ABSTRACT FROM AUTHOR]

Objectives: Traffic congestion detection is one of the key points and difficulties of urban traffic management. The existing congestion detection methods are based on road sections, which is not conducive to the extraction of spatiotemporal evolution information of congestion. Moreover, most of the detection only involves the degree of congestion but lacks the congestion type identification. With the classification and regression tree (CART) algorithm, this paper proposes a method for the classification and detection of traffic congestion points, which takes the road section point as the detection unit. In the practical application of this method, congestion points and their categories can be detected in real time according to the average running speed on the road section. Methods: Firstly, the road section is divided at a specific interval and mapped to be road section points. According to the abnormal rules and patterns of spatiotemporal road conditions, the spatiotemporal evolution patterns of four congestion types are analyzed with road section points as units. Secondly, the spatial and temporal sequence of road conditions of road section points is extracted on the basis of the road condition detection of road sections, and the spatial and temporal sequence of road conditions is classified and labeled according to different congestion types. Thirdly, four speed indexes are selected to constitute the attribute set of samples, and the speed of each road section point at each period is extracted according to the attribute set, which forms the dataset of decision tree learning. Finally, with the CART algorithm, the optimal model is obtained by the training with cross-validation to achieve the best generalization ability. Results: This paper proposes a classification and detection method of traffic congestion points based on CART. On one hand, congestion point detection is added to refine the basic unit of congestion detection, and on the other hand, congestion point type detection is also involved. Classification and detection of congestion points is helpful to improve the efficiency of traffic management. Conclusions: The proposed method is compared with the support vector machine classification model, and the experimental results show that the method in this paper has higher accuracy, higher recall rate, and better classification and detection timeliness. [ABSTRACT FROM AUTHOR]

In view of the slow detection speed of traditional grain flour species, this study implemented the random forest algorithm-assisted microwave nondestructive testing technology to identify grain flour species efficiently and accurately based on ZYNQ platform. By analyzing the hardware implementation of the random forest model, this paper proposed an improved model parameter structure, which could effectively save the consumption of hardware storage resources. In order to shorten the prediction time of the algorithm and reduce the power consumption of the system, it introduced the early termination recognition mechanism in the hardware implementation to avoid the unnecessary decision tree prediction process on the premise of keeping the accuracy unchanged. In view of the Zedboard development board, it designed a storage scheme for model parameters to make full use of on-board resources and ensure the normal operation of the system. The experimental results show that the implementation time of the scheme with ZYNQ is about 54. 2% shorter than that of the random forest algorithm with traditional CPU, and no loss of recognition accuracy is caused. [ABSTRACT FROM AUTHOR]

In order to improve the utilization rate of materials, a rebar cutting optimum model used integer programming and gradient boosting decision tree was established considering the influence of processing equipment on one-dimensional rebar products. On that basis, a product sorting optimum model used greedy algorithm was established combined with the intelligent sorting device independently developed by research group, it could promote rebar cutting optimum application in production and improve production efficiency. A case study shows that the presented cutting optimum method has more advantages than the traditional optimal methods in convergence performance, searching effectiveness and solution stability, and the utilization rate of materials has been effectively improved. The sorting optimization reduces the work intensity of field personnel, and greatly improve sorting efficiency. The presented method can guarantee the cost, quality, and efficiency of process, and provides a new way for one-dimensional rebar production in construction. [ABSTRACT FROM AUTHOR]

An accurate and real-time prediction has been urgently required for the pressure increment performance of centrifugal pumps under gas-liquid conditions during transportation. However, there is no appropriate prediction model for the pressure increment of the centrifugal pump at present, due to the complex gas-liquid flow. Machine learning can be widely expected to serve as a new idea for the performance prediction of centrifugal pumps. In this study, a machine learning model and database were established to investigate the effects of the inlet gas volume fraction, the rotational speed, and the liquid flow rate on the two-phase pressure increment performance of the pump. The results show that the inlet gas volume fraction was a direct factor for the deterioration of pump performance. The pump rotational speed and the liquid flow rate were greatly contributed to the improved performance of pump pressure increment. Three parameters were employed as the input features. The 234 sets of sample points from the experiment were divided into the training set (187 sample points) and test set (47 sample points) with the ratio of 8:2. Four models were constructed after pre-processing the data using Linear Regression (LR), BP Neural Network (BPNN), Support Vector Machine (SVM), and Random Forest (RF). The minimum Mean Square Error (MSE) was used to determine the optimal higher-order term regression function of the LR model, the optimal number of iterations, and the number of hidden layer neurons in BPNN. The grid search and 10-fold cross-validation were used to determine the optimal penalty coefficients and kernel parameters of the SVM, the optimal number of decision trees, and the minimum number of samples on leaf nodes in RF. Then, the Mean Relative Error (MRE), the Root Mean Square Error (RMSE), the determination coefficient (R²), and the confidence level were introduced as the evaluation indexes. The prediction ability of different models was evaluated on the training, the test, and the total data set. The results show that the RF model presented the best ability to predict the pressure increment of the pump, followed by the SVM, the BPNN, and the LR model. In the total data set, the MRE, RMSE, and R² of the predicted pressure increment by RF model were 3.51%, 1.06 kPa, and 0.993, respectively, and the relative error was no more than ±15% under the confidence level of 94.44%. The LR, BPNN, and SVM model failed to represent the characteristics of pump pressure increment, while the RF model performed better to characterize this trend near the pump surging region. It also needed to improve the pressure increment prediction of the RF model around the surging region under a small liquid flow rate. In addition, the importance of input parameters was also evaluated on the pressure increment by RF. Consequently, the rotational speed was the most important parameter on the pressure increment with a value of 52.67%, followed by the inlet gas volume fraction, and the liquid flow rate with 45.41% and 1.92%, respectively. Therefore, the gas-liquid two-phase performance prediction model using RF can be widely expected to better predict the two-phase pressure increment of centrifugal pump. The finding can also provide a strong reference for the design and selection of the centrifugal pump. [ABSTRACT FROM AUTHOR]

This paper applied the classical ID3 decision tree algorithm for discrete data classification. However, it required data discretization for continuous data processing, and this process easily caused information loss. This paper proposed the neighborhood equivalence relationship to induce the neighborhood ID3 ( NID3) decision tree algorithm. The NID3 algorithm improved the ID3 decision tree algorithm, which could directly implement continuous prediction and obtained better classification results. In the neighborhood decision system, this paper mined a neighborhood equivalence relationship. Then, based on the equivalent granulation of the neighborhood, this paper constructed the neighborhood information metric. Finally, this paper designed the NID3 decision tree algorithm based on the neighborhood information gain. As verified by both case analyses and data experiments, algorithm NID3 is effective for continuous data classification, and it outperforms ID3 algorithm in classification machine learning. [ABSTRACT FROM AUTHOR]

Objectives: The purpose of this study was to explore characteristics related to ambient air pollution health literacy (AAPHL) and to identify groups with lower AAPHL. Methods: A national, cross-sectional study of Taiwanese adults aged 20-70 years was conducted. Data were collected from September 11 to October 22, 2020. We conducted the survey through computerassisted telephone interviews. In addition, purposive sampling was used in online surveys. A 32-item Chinese version of the AAPHL assessment tool was used in telephone interviews and online surveys. Decision tree modeling was used to ascertain the characteristics of those with low AAPHL. Results: A total of 1,297 telephone interviews and 291 online surveys were completed, and 1,588 questionnaires were analyzed. The data indicated that having high school education, being unmarried, and living in southern Taiwan were associated with low AAPHL. Conclusions: The education level, area of residence, and marital status were the key risk factors affecting AAPHL. People with the identified characteristics should be regarded as the target population for health education interventions to enhance AAPHL levels in the general population and to reduce the impact of ambient air pollution on public health. [ABSTRACT FROM AUTHOR]

The winter crop has been one of the important crop types in China. Accurate and timely spatio-temporal distribution of planting area directly determines the grain output and economy, as well as the national food security. Taking the Huaihe River Basin as an example, this study aims to extract the planting areas of winter crops according to the phenology period using the Google Earth Engine cloud platform and the fusion of Landsat-7/8 and Sentinel-2A/B images. Firstly, a dataset of time-series images was constructed with a spatial resolution of 30m. A CFMask algorithm was selected to preprocess the images, thereby calculating the Normalized Difference Vegetation Index (NDVI). More importantly, the maximum NDVI of all high-quality images within 10 days was used to obtain time-series data with equal time intervals. The linear interpolation was utilized to fill the pixels without high-quality images. Savitzky-Golay (S-G) filtering (a second-order filter with a moving window of 9 observations) was adopted to smooth the NDVI time series for the removal of noise. As such, a smoothed NDVI time series was obtained with a 10-day interval. Secondly, the peak growth, sowing, and harvest periods were determined to select sample points of winter crops with different spatial distributions, according to the NDVI time series. Subsequently, the winter crops were sowed in mid-late October, when the NDVI values were the lowest. The NDVI values gradually increased, after the emergence of seedlings in early November. The crops stopped growing in January during the overwintering period, where the NDVI stayed the same over the whole period. Furthermore, the NDVI resumed growing and gradually reached the peak growth period, when the winter crops turned green in February. After that, the NDVI reached the peak at the heading stage, and then gradually decreased. Correspondingly, the NDVI dropped to the bottom, when the harvest was over from the end of May to June. According to these characteristics in the process of winter crops growth, the peak growth period was determined from March 20, 2018, to April 20, 2018, the sowing period was determined from October 11, 2017, to November 10, 2018, and the harvest period was determined from May 20, 2018, to June 30, 2018. Particularly, the maximum NDVI was achieved in the peak growth period and the minimum and median of NDVI in the sowing and harvest period. Finally, the classification model of a decision tree was constructed, according to the NDVI boxplots of winter crops and non-winter crops at different time periods. The planting area map of winter crops was also generated for the Huaihe River Basin. The results showed that the planting area of winter crops was 8.762×106 hm² in the Huaihe River Basin in 2018. Specifically, the user accuracy was 0.926, the producer accuracy was 0.970, the total accuracy was 0.958, and the Kappa coefficient was 0.912. Consequently, the large-scale planting area of winter crops was extracted accurately for the decision-making in similar areas. [ABSTRACT FROM AUTHOR]

In order to further improve the prediction accuracy of dissolved oxygen in aquaculture, this paper proposes a dissolved oxygen prediction model based on cross-validation grid optimization GBDTLSTM. First, the Gradient Boosting Decision Tree ( GBDT) is used to select characteristic factors that have a high degree of influence on dissolved oxygen. Then the Long Short-Term Memory Network (LSTM) is built based on the Keras deep learning framework, and the cross-validation grid optimization algorithm is used to optimize the LSTM parameters. Finally, the model in this paper is applied to a standard pond in the fishery base of Jintan City, Jiangsu Province to pr edict dissolved oxygen. The experimental results show that the mean square root error (RMSE), mean absolute error (M AE ) and mean relative error mean (MAPE) of the model are 0.208,0.158 and 2.635, respectively. The evaluation indexes are better than other comparison prediction models. It shows that the model has good predictive ability and generalization ability, which can meet the actual needs of modern aquaculture. [ABSTRACT FROM AUTHOR]

BKT can track the state of the learners ' knowledge, as well as predict their mastery level and future performance in the intelligent teaching system. Because BKT easily ignores the phenomenon of memory forgetting, and does not consider the impact of learning behavior on performance results, the model prediction results deviate from the actual situation. So this paper proposed a new Bayesian knowledge tracking model that integrated learner behavior and forgetting factors. Firstly, the method used the decision tree algorithm to process learning behavior data and introduced behavior nodes. Then, the model initialized the forgetting parameters and assigned them, and updated the algorithm of the learner ' s knowledge mastery level. Finally, the experiment compared the prediction accuracy of related models through the public data set provided by ASSISTMENTS. Experimental results show that the BF-BKT model can achieve better prediction accuracy. [ABSTRACT FROM AUTHOR]

It is a very meaningful direction to study data stream ensemble classification based on the condition of supervised or semi-supervised learning. This paper introduced three aspects including base classifiers, key technologies and ensemble strategies. The base classifiers mainly introduced decision trees, neural networks, support vector machines, etc. The key technologies were introduced from incremental and online aspects, and the ensemble strategies mainly introduced boosting, stacking, etc. This paper summarized and analyzed the advantages and disadvantages of different ensemble methods, comparison algorithms and experimental data sets. Finally, it gave the further research directions, including the handling of concept drift based on supervised and semi-supervised learning, the study of homogeneous integration and heterogeneous integration, and the classification of data stream ensemble based on unsupervised learning. [ABSTRACT FROM AUTHOR]

[Objective] Existing in-depth research on tea yield prediction is lacking, so the present paper aimed to analyze the influence of different features or different tea areas on the prediction results, and determine the decisive and auxiliary factors that affect the tea field, thus to provide guidance for the management of tea production. [Method] Zhejiang province was taken as a research area. According to the daily ground observation meteorological data, the tea statistical yield data and social development factors in Zhejiang province from 1995 to 2016, a total of 94 meteorological features such as temperature and humidity, sunshine, precipitation in each year and each season. Three spatial features of the major tea producing areas, latitude, longitude, elevation, the four social development features, tea garden area, total power of agricultural machinery, effective irrigation area, rural electricity consumption were extracted and the actual annual yield of each major tea producing area was taken as the target variable. Then regression models of tea yield prediction by the gradient boosting decision tree with combining multiple feature variables were constructed, and the relationship between each feature and yield was analyzed according to the importance of the feature. [Result] (i) Compared with using only social development features or space meteorological features, the prediction model combining social development features and space meteorological features was better. Its decision coefficient reached 0. 90, root mean square error was 1492 t, mean absolute error was 1050 t. In addition, social development features played a significant role in the tea prediction; (ii) Compared with the summer and autumn, the prediction accuracy of tea yield in spring and winter was higher, and the values of determination coefficient reached 0. 89; (iii) In addition to the social development features, among the space meteorological features, the spatial features, annual meteorological features, spring meteorological features, and winter meteorological features had a great influence on the yield. [Conclusion] The tea yield prediction model proposed in this paper has important reference value for tea yield prediction and analysis of influencing factors. [ABSTRACT FROM AUTHOR]

OBJECTIVE: To probe into the economics of Tongluo Kaibi tablets, Biqi capsules and Gulong capsules combined with methotrexate in the treatment of rheumatoid arthritis, so as to provide reference for clinical drug selection and decision-making by health and medical insurance departments. METHODS: Based on the health system perspective, a three-month decision tree model was constructed by using Treeage Pro software, and cost-effectiveness analysis of three kinds of traditional Chinese patent medicines combined with methotrexate in the treatment of rheumatoid arthritis was performed. Efficacy data and adverse drug reactions data were derived from published randomized controlled trials. The incremental cost-effectiveness ratio ( ICER) was used as the evaluation index and sensitivity analysis was carried out. RESULTS: Tongluo Kaibi tablets combined with methotrexate had the lowest treatment cost and the best effect, ICER<0. The results of sensitivity analysis showed that the most influential factors on the results were the total effective rate and the drug price of Tongluo Kaibi tablets. When willing to pay(WTP) was from 0 and 2 000 yuan, the acceptable probability of patients in the Tongluo Kaibi tablets group was higher than that in the Gulong capsules group; when WTP was > 18 yuan, the acceptable probability of patients in theTongluo Kaibi tablets group was higher than that in the Biqi capsules group. CONCLUSIONS: Among the regimens of Tongluo Kaibi tablets, Biqi capsules and Gulong capsules combined with methotrexate in the treatment of rheumatoid arthritis, Tongluo Kaibi tablets combined with methotrexate has the most cost-effectiveness advantage, followed by Biqi capsules, and Gulong capsules has the least economy. [ABSTRACT FROM AUTHOR]

[Objectives] An automatic method for freshwater fish species classification was proposed in this paper, based on the fish images collected by using a homemade fish image acquisition device. [Methods] A self-template matching method was developed to segment the fish body region from the binary image, which was obtained after the image pre-processing operations. Then the shape and texture features of the fish body were extracted, where the latter were obtained using three methods, including gray level-gradient co-occurrence matrix(GLGCM),local binary pattern(LBP),and Gabor filter. Principal component analysis(PCA) was carried out to transform a high dimensional feature set of GLGCM texture and shape, LBP texture and shape, or Gabor texture and shape into smaller ones that contained at least 85% of the information in the large feature set. 70% and 30% of the feature sets were used to train and validate species classification models established using seven machine learning methods[Naive Bayes, K-neares neighbors(KNN),linear regression, randow forest, decision tree, support vector machine(SVM),gradient boosting decision tree(GBDT)]. [Results] Fish body segmentation test indicated that the developed self-template matching method could segment the region on interest with a correct rate of 99.79%,which was achieved without the need of an extra fish body template library. Classification models validation results indicated that the random forest classifier using reduced dimensional GLGCM texture and shape feature performed best, and an average accuracy of 99.52% was achieved when it was used to classify all the 14 freshwater fish species. [Conclusions] The self-template matching method developed in this study could extract fish body region well without the need of an extra fish body template library. The reduced dimensional GLGCM texture and shape feature set was suitable for identifying the species of freshwater fish. [ABSTRACT FROM AUTHOR]

Current clustering methods study data clustering problems only from one angle, it is insufficient to considerate clustering chaotic big data of Internet of Things based on cloud pattern with low clustering quality. To achieve agile, intelligent and stable clustering on big data of Internet of Things, with studying general cloud pattern description models on events of Internet of Things, general cloud pattern analysis models on chaotic correlation features of events of Internet of Things, extracting algorithms on chaotic correlation features of events of Internet of Things based on cloud pattern, correlation mining of big data of Internet of Things based on cloud pattern chaotic correlation features, improved decompositing singular value algorithms, grid coupling clustering algorithms, K-means algorithms, decision tree learning methods, methods of analysis principal components, stratification merging methods and distribution probability function, this paper designed an agile, intelligent and stable clustering algorithm on big data of Internet of Things based on chaotic correlation features of events. Finally, it carried out validating experiments, and compared performance of this proposed algorithm with traditional algorithms. Experimental results show this algorithm has shorter clustering time, less error and higher agility, has better intelligence, dynamic evolution, stability than those of traditional algorithms. Therefore, this proposed algorithm achieves effective clustering on big data of events of Internet of Things with chaotic correlation features based on cloud patterns, has higher utility. [ABSTRACT FROM AUTHOR]

Due to the high-speed requirement of high-end network devices, hardware using TCAMs has been the dominant implementation of packet classification in industry. Despite its capability for line-speed queries, TCAM is not only power hungry but also capacity inefficient. By making use of a pre-classifier to activate TCAM blocks selectively, many research efforts significantly reduce the power consumption of TCAM. However, these bottom-up based pre-classifiers achieve power savings at the expense of poor utilization of TCAM capacity, and the potential of power reduction is not fully exploited in many cases. This paper proposed a power-saving pre-classifier for TCAM based packet classification, which constructed based on decision trees. By grouping rules with respect to their small fields, rules could be recursively mapped into decision trees without the trouble of rule replications, so that a top-down traversal algorithm could be well applied for obtaining index items. Experimental results show that for rule sets up to one hundred thousand entries, the proposed design achieves 98. 2% power reduction with a TCAM storage overhead of 1. 3% on average. [ABSTRACT FROM AUTHOR]

Heilongjiang province is the main area for paddy cultivation in China, and the phenomenon of paddy field expansion has contributed to huge changes in the land types in agricultural areas. Remote sensing is employed to rapidly and dynamically monitor the spatial and temporal changes of paddy fields, thus providing scientific support and decision-making basis for rational cultivation of crops and exploitation of land resources. Based on the above, the Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance and vegetation index data sets were selected as the main data source and the Landsat data set was chosen as the auxiliary data source in the present study. As MODIS images have the characteristics of large width and high update frequency, it is an ideal tool for accurate identification of large area crops. The current study remotely decoded paddy fields, drylands, river beaches, swampy meadows, forests, water, and towns in Heilongjiang province from 2003 to 2018 based on the decision tree model. Besides, the data from 2003 to 2010 was the calibration group, and the data from 2011 to 2018 was the validation group. Since the phenological characteristics and exponential intervals of the land classes all showed the difference, the classification rules of the land classes were also different. Statistical analysis was performed based on the spectral characteristics and time-series curves of the indices, including the Enhanced Vegetation Index (EVI), Normalized Difference Vegetation Index (NDVI), and Land Surface Water Index (LSWI). Meanwhile, the classification rules for each land class were presented as follows: forests were extracted by the EVI and slope data threshold method on April 6. Water was extracted by the NDVI threshold method on October 16. Supervised classification was used to extract towns from the LSWI time-series and wetlands from the NDVI time-series. After carrying out repeated experiments, NDVI, LSWI, and Band 6 were used to identify the paddy field, and the threshold conditions included 0.45-0.77, 0-0.56, and 120-1 530 nm, respectively. The classification result images were verified by high-resolution Landsat images and statistical almanac data, respectively. The Kappa coefficient of the 2003-2018 paddy fields identification reached 0.899-0.961, the overall classification accuracy reached 85.5%-92.3%, and the paddy fields matched the statistical almanac data. To compare the advantages and disadvantages of decision tree model construction, the maximum likelihood method was selected for the comparison. In terms of the control group, the maximum likelihood classification method was used to identify paddy fields under the condition that other land classification rules were unchanged. From 2003 to 2010, the accuracy of the maximum likelihood method was 0.643-0.756, which was significantly lower than that of the decision tree method from 0.923-0.961, indicating that the classification of paddy fields using the threshold method was more effective compared with the maximum likelihood method. The classification results suggested that the area of paddy fields in Heilongjiang province expanded 3 times from 2003 to 2018, and the center of gravity of paddy fields in the sowing area extended approximately 160 km to the north. Paddy field expansion increased linearly, with an average expansion of 158 100 hm² per year. From 2003 to 2018, the cumulative conversion from dry land was 2 502 400 hm², and 154 900 hm² of wetlands had been reclaimed in total. Moreover, the decision tree model proposed in the present study had provided an effective method for extracting paddy cultivated areas in Heilongjiang province, which could also offer lessons for land class identification in similar areas. [ABSTRACT FROM AUTHOR]

Most multivariate decision trees are applicable for only the numerical data. To solve the classification problem on categorical attribute data， an applicable multivariate decision tree(CMDT) algorithm is proposed. The center of the sample set on the categorical attributes， and the distance between the samples and the centers are defined with statistics for frequency distribution of categorical attribute values in each category or each cluster. Weighted k-means algorithm is utilized to split the nodes in the decision tree. The proposed multivariate decision tree is applicable for numerical data， categorical data， and mixed data. Experiment results show that the classification model based on the proposed algorithm can get more concise tree construction and higher generalization accuracy than that based on the classic decision tree algorithms with different kinds of data. [ABSTRACT FROM AUTHOR]

Aiming at the problem that the computational overhead of cloud storge data deduplication scheme called BL-MLE is too large, which was proposed by Chen et al., this paper improved the scheme, proposed a more efficient data deduplication scheme. Firstly, it analyzed the BL-MLE scheme, pointed out its shortcomings in computational efficiency. Then it used hash function and tag decision tree to improve the BL-MLE' s block tag generation process and block tag comparison process. Finally, this paper experimentally simulated the improved scheme. The results show that the improved scheme requires fewer times in block tag comparison,lower time overhead in block tag generation, can better adapt to the current cloud storage environment. [ABSTRACT FROM AUTHOR]

Aiming at the problems of large data redundancy and unclear characteristics of multi-sensor data acquisition in the body area network, this paper proposed a data fusion method based on deep neural decision forests( DNDF). The method used a convolutional neural network for feature extraction, and classified the data after placing the decision tree into the fully connected layer. The DNDF method could extract the critical features of the multi-dimensional data effectively, and preserved the correlation between the data better. The experiment used the AReM dataset. The results show that the DNDF method has better performance than other traditional algorithms and achieves higher classification accuracy. [ABSTRACT FROM AUTHOR]

In order to compare and analyze the network traffic classification effect of different machine learning algorithms in the software defined network (SDN) environment, the Moore dataset was balanced, and four classic machine learning algorithms including KNN, random forest (RF), support vector machine (SVM) and gradient lifting decision tree (GBDT) were supported on the machine learning platform RapidMiner to select different classification features for classification experiments. Experimental results showed that compared with the other three algorithms, the GBDT algorithm could obtain better classification results in a shorter time. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the recognition efficiency of sensitive word deformed bodies of the network text is not high, this paper proposed a sensitive word deformed bodies recognition algorithm based on decision tree. Firstly, it studied sensitive words and deformed bodies by analyzing the characteristics of Chinese characters and pronunciation and so on. Secondly, it constructed a sensitive word decision tree based on sensitive word library. Finally, it calculated the text sensitivity of new media such as Weibo by multi-factor improved model. The experimental results show that the proposed algorithm can achieve the highest recall rate and precision rate of 95% and 94% respectively when identifying Chinese sensitive words and deformed bodies. Compared with the improved algorithm based on the finite automaton, the recall rate and the precision rate are increased by 19. 8% and 21. 1% respectively. Compared with the sensitive information decision tree information filtering algorithm, the recall rate and the precision rate are increased by 17.9% and 18.1% respectively. The analysis show that the algorithm is effective in the recognition and automatic filtering of sensitive word deformed bodies. [ABSTRACT FROM AUTHOR]

Rubber forest has an increasingly important impact on the environment and social economy. The huge demand has created a demand-supply gap in the process of economic globalization and the rubber forest planting, especially that in the border areas of different countries has been widely concerned in various fields. In this study, the rubber forest distribution was extracted by using the cloud computing technology of Google Earth Engine platform and the integration of multiple Landsat OLI remote sensing images from 2015 to 2019 in the border areas of China, Laos and Myanmar. The rubber phenology characteristics were obtained through rubber time series analysis. The different feature parameters were selected, and the differences of each parameter in foliation and defoliation period were compared to distinguish rubber forest and other land coverage types. Then the classification model of expert knowledge decision tree was constructed based on the calculated segmentation threshold of each parameter, and the algorithm is applied to the whole research area of the border areas of China, Laos and Myanmar. The results showed that time series NDVI in February (defoliation period) and April (foliation period) of rubber in the study area had good performance to distinguish rubber forest and other land coverage types. The overall accuracy of extraction was 90.32% and Kappa coefficient was 0.87. Both the overall accuracy and Kappa coefficient met the accuracy requirements of general production. Compared with the former researches, the method based on Google Earth Engine using rubber phenology calculation parameters to extract rubber forest in a large research area has a high accuracy. The total area of rubber forest extracted was 126.29x104 hm², including 52.37x104, 56.93x104 and 16.99x104 hm² of rubber forest extracted from Xishuangbanna, Myanmar and the five northern provinces of Laos, respectively. It is also found that the areas of rubber forests were different in these regions because the different actual situation of Laos and Myanmar produce differential policies in the process of alternative policy development. The cloud computing technology based on Google Earth Engine platform can overcome the lack of computing power of large-scale rubber monitoring in time and space, and provide scientific basis and decision support for the rational rubber layout and regional sustainable development in the border areas of China, Laos and Myanmar. [ABSTRACT FROM AUTHOR]

To improve the accuracy of human behavior recognition and reduce the complexity of the implementation process, this paper raised a method using the data of acceleration sensor and gyroscoped in smartphone to recognize six kinds of daily basic behaviors. It collected and preprocessed data of acceleration sensor after analyzing the framework of sensor, then it used principal component analysis method combining with existing knowledge to reduce the dimension of the statistical characteristics of the original data. Then it used machine learning algorithm to classify and recognize the behavioral characteristics, which aimed at simplifying the basic behavior recognition process and improving the utilization rate of the data. The experimental result shows that the combination of decision tree and support vector machine classifier is effectual, and the recognition accuracy can be close to 97% . [ABSTRACT FROM AUTHOR]

Aiming at the problem of low recognition accuracy of rotating machinery faults, this paper proposed a rotor system fault identification method based on the combination of EEMD with energy moment and NRS. Firstly, this method used the EEMD to decompose the non-stationary vibration signal into several stable IMF and calculated the energy moment of the IMF component. This energy moment was used as the condition attribute to describe the fault state to establish the fault identification decision table. Then this paper used the neighborhood rough set to perform attribute reduction on the decision table to eliminate the redundant attribute. Finally, it used the reduced sensitivity feature subsets as input into decision tree C4.5 algorithm for recognition. The experimental results of fault feature set verify effectiveness of this method. [ABSTRACT FROM AUTHOR]

In order to effectively extract the key features from the extremely short texts, this paper proposed an extremely short text classification model based on SVM. Firstly, by the data cleansing on the original data, the cleaned data was processed by the jieba segmentation and TF-IDF. Then the 1-0 test verified the validity of the model. Finally, 9906 pieces of extremely short texts in Wuhu city community management platform were used as the sample in this experiment. The results show that the proposed method can effectively improve classification accuracy compared to other traditional methods, such as naive Bayes, logistic regression and decision tree. At the same time, the matching results in terms of misclassification and accuracy are more balanced. [ABSTRACT FROM AUTHOR]

Accurate detection of historical status of granary is helpful for its management. Based on statistical analysis of grain temperature in granary, this paper presents a method to detect the storage status of the granary in attempts to resolve the problem of time-consuming and tediousness faced by inventory inspection. Warehouse with state of being empty, filled with fresh grain and ventilation was used as an example. Analysis of the distribution of statistical characteristics of historical grain temperature showed that the empty warehouse could be detected using the difference in temperature between the upper and lower adjacent layers of the grain bulk as well as the variation of the grain temperature; the fresh grain could be detected by the difference in temperature between adjacent layers and the difference in standard deviation of the grain temperature; the ventilation could be detected by the change in the grain temperature and the standard deviation difference of the grain temperature. The threshold intervals of the characteristic parameters of the three states were set preliminarily from analysis of the statistical characteristics of the grain temperature. Based on the threshold intervals, three optimal levels were selected and the orthogonal experiments of two-factors and three-levels were designed. In the same grain depot with 11 granaries, new and ventilated granaries were chosen for the orthogonal experiments. The optimal threshold range of the statistical characteristic parameters were determined by analyzing the statistical characteristics of the historical grain temperature changes in the 11 granaries. Multi-variable decision tree for inventory status detection was designed and the optimal threshold of the characteristic parameters of the multi-variable decision tree were as follows. The temperature difference between adjacent layers for empty granary status detection was [-0.12, 0.12] ℃. The temperature difference between adjacent layers was more than 0.02 (or 0.03) ℃, and the standard deviation was less than 2.0 (or 2.1) ℃. The average temperature change rate in the parameters for detecting the ventilation characteristics was [-0.5, 0.5] ℃/d and the standard deviation change rate was [-0.21, 0.21] ℃/d . Finally, seven granaries in different provinces were selected to test the detection method. The results showed that the accuracy and recall rates were 78% and 82% respectively for the empty granary state, and 74% and 70% respectively for new grain state, and 91% and 88% respectively for ventilated state. The test results showed that the proposed method detected granary at empty and ventilated state more accurate than at new grain state. [ABSTRACT FROM AUTHOR]

The retrieval system uses learning to rank ( LtR) algorithms to produce a ranking model from public available training sets. The time required reducing to retrieve data is an important research direction of the retrieval system. In order to reduce the retrieval time, this paper studied the pruning strategy of ranking model and cache. Using the redundancy characteristics of the decision tree and the cache, it proposed a pruning decision tree model and block-wise algorithm. Finally, this paper aimed to answer the question that whether it could improve the efficiency of the ranking model without affecting the effectiveness of model. Experiments on two publicly dataset show that the pruning decision tree model and block-wise algorithm can effectively reduce the scoring time per query. [ABSTRACT FROM AUTHOR]

Network entity inference is one of the important contents in cyberspace surveying and mapping. Network entity inference and calibration need to synthesize multi⁃source data，classify nodes in network space by synthetical judgment. Thus，the entity classification model in network space is proposed. Based on this model，a low ⁃overhead network entity detection classification method is proposed. Firstly，for IP addresses detected by detection，the alias parsing technology is used to map multiple IP addresses belonging to a device into a network entity；Then，the decision tree is used to classify network entities in a coarse⁃grained manner； Finally，the Bayesian network is used to classify them in detail. Taking a city in Jiangsu Province as an example，the detection and analysis are carried out and compared with the recorded data. The experimental results show that the method can effectively classify various types of entities in the network space，thus providing support for network space map construction，situation analysis and other applications. [ABSTRACT FROM AUTHOR]

In order to make the assistant decision of airport pavement management more accurate and meet the requirements of intelligence，the existing network-level decision making algorithms need to be improved. Based on the concept of information entropy，we present a decision tree model of civil airport cement pavement maintenance plan. The model test and evaluation results show that the model has high accuracy， cross-validation and self-correcting ability. We take Shanghai Pudong International Airport pavement management system as an example. Compared with traditional models，the decision tree model is proved to encompass multi-factors， like pavement performance， structural performance and subjective factors of management needs. The results demosnstrate that the proposed model is more fit the engineering in practice. [ABSTRACT FROM AUTHOR]

In order to solve the problem that the use of softmax as a long-short-term memory network classifier leads to the lack of generalization ability of the entity relationship recognition model, it is not suitable for the extraction of TCM entity relationships. This paper proposed a bidirectional long short-term memory (BILSTM) relational identification algorithm (BILSTM-GBDT) that incorporates a gradient boosting decision tree (GBDT) Firstly, it trained the Chinese medicine text vector by word2vec, then extracted the high-order features by the bidirectional long short-term memory network based on the attention mechanism. Finally, it used the integrated classification model gradient lifting tree as the feature classifier to improve the relationship recognition effect. Experimental results on multiple relational corpora such as Chinese medicine show that the model has higher accuracy, recall and F value than traditional SVM method, GBDT method and deep learning method. [ABSTRACT FROM AUTHOR]

It is of great significance to monitor dynamic change of wetland over the Yellow River Delta for rational utilization, development and protection of wetland resources. Both Gaofen-3 (GF-3) SAR data and Sentinel-2B multispectral data were used to analyze the spectral, index, polarization scatter and texture feature information of seven types of ground objects over the Yellow River Delta wetland, and then supervised classification was implemented with maximum likelihood (ML), decision tree (DT) and support vector machine (SVM) classifier. The performances of both the joint and the individual classifications with GF-3 and Sentinel-2B data were also evaluated. The results of three algorithms show that the overall accuracy of the joint classification can reach 90.4%, 95.4%, 95.7%, significantly higher than that of the individual classifications, showing the promising potential of GF-3 SAR and Sentinel-2B multi-spectral images in joint wetland classification. [ABSTRACT FROM AUTHOR]

The evaluation of subhealth of traditional college teachers is lack of timeliness, objectivity and efficiency. In the big data environment, the subhealth assessment model can be established more effectively by information technology, so as to support the evaluation and prediction of subhealth state of university teachers. Firstly, according to the analysis of multidimensional subhealth impact factors of university teachers, the paper constructs a decision conceptual model of the impact of multidimensional subhealth, followed by the analysis and data processing of the sample data characteristics, based on the detailed process of using CART algorithm. Finally, the parallel implementation process and experimental verification based on Spark are given. The subhealth concept model of university teachers objectively reflects the evaluation factors of sub health. The decision tree model can support the prediction and analysis of subhealth of university teachers, and experiments verify the validity, immediacy and accuracy of the model. [ABSTRACT FROM AUTHOR]

For the satellite video motion detection, two issues are harmful to the subsequent application. One is the local parallax“pseudo-motion”, and the other is“Ghost pseudo-motion”, which may lead to serious error detection. For these two issues, a decision tree-supported local differentiated updating background subtraction method is proposed. Firstly, the initial background model is established. One other parameter, the updating factor, is added based on the ViBe model. Each pixel in the model has a unique dynamic updating factor. In the next place, we perform the pixel-wise motion detection based on the modified model. The detection results are segmented into targets by using the connected component analysis-labeling method. Afterwards, the decision tree model is established based on the target characteristics. And the targets can be weekly classified into different categories. In the end, the background model samples and the factor parameters are differentiated updated according to the categories results. Using SkySat and Jilin No. 1 video data to do experiments. The experimental results show that our proposed method has favourable effect and performance, with the recall rate of results superior to 0.909. Comparing with the classical ViBe method, the error detection removal rate is superior to 90%. [ABSTRACT FROM AUTHOR]

The color between stalks and leaves of tea in summer and autumn is similar, which means the traditional color sorter is difficult to sort based on optical characteristics. To realize the rapid modeling of tea selection algorithm and improve the sorting accuracy, a method for sorting the fine and bad products of tea by multi-feature vectors based on the morphological characteristics was introduced in this paper. First, Wuyishan Dahongpao tea was selected as a test sample to collect images during the dynamic drop process. The blue element image was extracted, and single sample’s binary image and edge were obtained by analysis of whole image connection area. Then, feature extraction program was developed based on image processing algorithm to extract morphological feature parameters of the tea samples automatically. Four simple shape descriptors-the sample perimeter, area, the length and width of minimum bounding rectangle were extracted. On this basis, eight complex shape descriptors-circularity, rectangularity, linearity, slightness, diameter, diagonal of minimum bounding rectangle, compactness and centroid were calculated. In addition, the random forest algorithm was used to determine the above features weight, the feature was selected according to weight threshold. Finally, logistic regression (LR), decision tree (DT) and support vector machine (SVM) that three different classification algorithms were established to classify the samples, verify the validity of the features and analyze the effects of different classification algorithms on the classification of tea. The original data were normalized and randomly segmented 80% used for training, 20% for testing. 10-fold cross-validation was used to select the optimal parameters of the classification model, and the training dataset was randomly divided into 10 parts, of which 9 parts were used for training, and the remaining 1 part was used for verification. According to the above machine learning system parameter optimization process to obtain the logical regression, decision tree and support vector machine optimal model, and statistical the final evaluation results on test dataset. The test results showed that: 1) The circularity weight was the highest, at 0.467, and five eigenvectors of circularity, rectangularity, linearity, perimeter and compactness were finally selected with the weight threshold value which was 0.05; 2) In the test dataset, the average accuracy of the three classification algorithms was 0.924, suggesting that the established tea morphological feature descriptors has certain separability and better effect; 3)When testing test-dataset, the accuracy score was 91.7% and F1 score of logistic regression (LR) was 92.9, the accuracy score was 91.7% and F1 score of support vector machine (SVM) was 94.7%.Support vector machine (SVM) algorithm was the best recognition effect in three classification algorithms; 4) From three different classification algorithms assessment score deviation, we can see that the generalization ability of the logic regression algorithm was stronger, the decision tree algorithm has a greater risk of over fitting. We get the lowest accuracy and F1 score of the logistic regression algorithm, while the support vector machine accuracy and F1 score were the highest, so in the evaluation of eigenvector comparability, multiple algorithms can be selected to evaluate the results of the average as the final basis for evaluation. In the experiment, we acquired dynamic image, which stay in line with the actual working conditions of the tea selection process, and can be extended to the actual processing of tea production. [ABSTRACT FROM AUTHOR]

Based on multisensory data? including sptical and thermal images from MODIS» and dualpolarization SAR image from RadarSat-2, sea ice in Beaufort Sea is classified, sea ice leads are identified and accuracy of the result is also evaluated. Experiment shows that thermal images from MODIS is only useful in extracting lead area with coarse resolution, while the high-resolution SAR image could provide more information on sea ice type in lead. However, overlay of signals from different ice type largely reduces the overall accuracy from supervised classification of RadarSat-2 image. Therefore ,decision tree is established to incorporate multisensory data. Characteristics of different ice types and leads is analyzed and utilized in decision tree to detect ice leads in different stages of development. Validation shows that overall accuracy of the result from decision tree is 14. S% higher than that from supervised classification. The sequential structure composed of various development stages of ice leads is confirmed in Sentinel-2 images. The result might facilitate accurate calculation of heat flux and ice production, as well as ship navigation with detail ice type in lead [ABSTRACT FROM AUTHOR]