Original Data Points Research Articles

On coresets for support vector machines

We present an efficient coreset construction algorithm for large-scale Support Vector Machine (SVM) training in Big Data and streaming applications. A coreset is a small, representative subset of the original data points such that a model trained on the coreset is provably competitive with that trained on the original data set. Since the size of the coreset is generally much smaller than the original set, our preprocess-then-train scheme has potential to lead to significant speedups when training SVM models. We prove lower and upper bounds on the size of the coreset required to obtain small data summaries for the SVM problem. As a corollary, we show that our algorithm can be used to extend the applicability of any off-the-shelf SVM solver to streaming, distributed, and dynamic data settings. We evaluate the performance of our algorithm on real-world and synthetic data sets. Our experimental results reaffirm the favorable theoretical properties of our algorithm and demonstrate its practical effectiveness in accelerating SVM training.

Analysis and Comparison of Quasi-Elastic Neutron Scattering (QENS) Spectra Model

The Quasi-Elastic Neutron Scattering (QENS) experiment is a novel method to study confined water dynamics by measuring the neutron energy transfer ( E )and scattering vector (Q) after the neutron interacts with the sample target. By adopting a suitable theoretical model and performing nonlinear least-squares fitting analysis on the Quasi- Elastic Neutron scattering spectra, the relevant physical parameters describing the water dynamics can be obtained, and then the constrained water dynamics in the sample can be detected. There are many types of Quasi-Elastic Neutron Scattering spectra fitting analysis models, and the fitting effects of different models and the relevant physical parameters obtained to characterize water dynamic information are not the same. This paper uses the Jump-diffusion and Rotating-diffusion Model (JRM) and the revised Jump-diffusion and Rotating-diffusion Model (rJRM) to fit the QENS spectra of pure magnesium silicate (M-S-H) and pure calcium silicate (C-S-H) samples, respectively. The measurement temperatures of QENS spectra from M-S-H and C-S-H samples are 210K up to 280K and 230K up to 280K, respectively, the E ranges from -120 to 120μev and the Q value 0.3 A^(-1)~1.9 A^(-1). Different physical parameters describing the dynamic information of confined water can be obtained by using the two models for QENS spectra fitting analysis. Compared with the JRM model, the physical parameters obtained by using the rJRM model are more comprehensive, and it is suitable for all neutron energy transfer and scattering vector measurements. From the qualitative point of view, the QENS spectra fitting curve of rJRM model is more consistent with the original data points. From the quantitative point of view, the chi-square value fitted by the rJRM model is more consistent with the mathematical statistics law than the JRM model, and the time required for nonlinear fitting is shorter.

Proposal of Unsupervised Gas Classification by Multimode Microresonator

A high-accuracy unsupervised classification model is developed in a multimode self-interference microring resonator (SIMRR). For the SIMRR, there are many whispering gallery modes (WGMs) present. Each of these resonance modes supported by the SIMRR has a different response to different target parameters, so that the SIMRR-based sensor has the super capability to distinguish between multiple components. In the classification model, principal component analysis (PCA) is firstly used to reduce the dimensionality of this multimode sensing information from the SIMRR-based sensor. When the original higher-dimensional data points are projected onto the lower-dimensional data with only the first few principal components, they can be easily categorized into several different types by using density-based spatial clustering of application with noise (DBSCAN) algorithm. As an example, the unsupervised classification method is numerically validated based on a designed three-gas SIMRR-based sensor. The numerical results prove that the classification model can achieve an ultra high classification accuracy for the designed three-gas sensor with more than 60 dB in signal-to-noise ratio.

Self-adaptive trajectory prediction for improving traffic safety in cloud-edge based transportation systems

Intelligent transportation brings huge benefits to humans’ life and Industrial production in terms of vehicle control and traffic management. Now, the development of edge-cloud computing has once again promoted intelligent transportation into a new era. However, the development of intelligent transportation inevitably produces a large amount of data, which brings new challenges to data privacy protection and security. In this paper, we propose to develop an improved trajectory prediction framework based on the self-adaptive trajectory prediction model (SATP), which could significantly enhance traffic safety in transportation systems. The proposed framework is capable of guaranteeing the accurate trajectory prediction of moving target under different application scenarios. In particular, to reduce the size of original trajectory point data collected by sensors, the angle change and minimum description length (MDL) principle are first combined to remove the redundant points in raw trajectories. The obtained points can then be reduced for model using the two-step clustering method. To further enhance the prediction performance, we add the “self-transfer” to the original model to solve the problems that the state of original SATP model may be discontinuous. Furthermore, we propose to develop a trajectory complementation method based on Bezier curve to improve the prediction accuracy. Finally, by comparing the two-step clustering method with the commonly-used SinglePass and density-based clustering method (DBCM) algorithms, the proposed two-step clustering policy greatly reduce the time cost of clustering. At the same time, by comparing the improved SATP model with the original model, the results show that the improved SATP method can greatly improve the speed of prediction model.

A Bayesian Framework for Large-Scale Identification of Nonlinear Hybrid Systems

Abstract In this paper, a two-level Bayesian framework is proposed for the identification of nonlinear hybrid systems from large data sets by embedding it in a four-stage procedure. At the first stage, feature vector selection techniques are used to generate a reduced-size set from the given training data set. The resulting data set then is used to identify the hybrid system using a Bayesian method, where the objective is to assign each data point to a corresponding sub-mode of the hybrid model. At the third stage, this data assignment is used to train a Bayesian classifier to separate the original data set and determine the corresponding sub-mode for all the original data points. Finally, once every data point is assigned to a sub-mode, a Bayesian estimator is used to estimate a regressor for each sub-system independently. The proposed method tested on three case studies.

Role of Human Body Posture Recognition Method Based on Wireless Network Kinect in Line Dance Aerobics and Gymnastics Training

With the rapid development of the information society, human body gesture recognition has become an important technology for human‐computer interaction. This paper combines Kinect’s human bone monitoring technology with auxiliary gymnastics training. The gymnastics and dance training can correct students’ wrong movements in time through feedback and improve the training efficiency, so as to meet the needs of nature and harmony of human‐machine interaction. In this paper, based on the wireless network Kinect, the human body posture recognition method and tracking technology are studied, and the joint point angle representation method based on the fixed axis is proposed, and the posture recognition method based on the joint point angle is improved, which can accurately recognize the human body posture. Aiming at the situation that the human joint points are occluded, the human joint point repair algorithm is improved. The algorithm is based on the proportion of human bone nodes and the characteristics of human motion, and based on geometric principles, it repairs the occluded points. The feasibility of the original joint point data, angle feature, and distance feature in expressing human behavior is analyzed through experiments, a standard gymnastics movement database is established, and new gymnastics movements can be entered at any time. A gymnastics auxiliary training system is designed, which can analyze and evaluate the exercises of the trainer from the joint point coordinates and the angle formed by the joints and provide the trainer with intuitive error correction prompts. The human body posture recognition method studied in this paper can accurately give the difference between the trainer’s movement and the standard movement, and the trainer can adjust the movement posture according to the prompts, improve the level of gymnastics, and achieve the purpose of auxiliary training. Experiments show that the algorithm model has an accuracy rate of 95.7% for human body posture recognition, and it plays a huge role in line dance aerobics and gymnastics training.

Optimal Feature Manipulation Attacks Against Linear Regression

In this paper, we investigate how to manipulate the coefficients obtained via linear regression by adding carefully designed poisoning data points to the dataset or modify the original data points. Given the energy budget, we first provide the closed-form solution of the optimal poisoning data point when our target is modifying one designated regression coefficient. We then extend the analysis to the more challenging scenario where the attacker aims to change one particular regression coefficient while making others to be changed as small as possible. For this scenario, we introduce a semidefinite relaxation method to design the best attack scheme. Finally, we study a more powerful adversary who can perform a rank-one modification on the feature matrix. We propose an alternating optimization method to find the optimal rank-one modification matrix. Numerical examples are provided to illustrate the analytical results obtained in this paper.

Room-Temperature Hydrogen- and Ammonia Gas-Sensing Characteristics of a GaN-Based Schottky Diode Synthesized With a Hybrid Surface Structure

A GaN-based Schottky diode synthesized with a hybrid surface structure is used to fabricate a new room-temperature (25 °C) hydrogen and ammonia gas sensor. This hybrid surface structure includes platinum (Pt) nanoparticle (NP)/SiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> nanosphere (NS) mixtures and a Pt thin film. The employed hybrid surface structure can effectively increase the catalytic reactivity of the Pt metal and enhance the related gas-sensing characteristics. In the experiment, a higher sensing response of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$7.3\times 10^{{5}}$ </tex-math></inline-formula> ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2.1\times 10^{{1}}{)}$ </tex-math></inline-formula> is obtained under 1% H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> /air (1000 ppm NH <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> /air) gas at 25 °C. Moreover, extremely low detecting levels of 100 ppb H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> /air and 100 ppb NH <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> /air are acquired at 25 °C. The related hydrogen- and ammonia-sensing mechanisms are elucidated in this work. For wireless transmission and Internet of Things (IoT) application, the Kalman filter and shape-preserving piecewise cubic interpolation (SPPCI) algorithms are employed to substantially reduce redundant data and restore original results. More than 86.7% (97%) of the original data points are removed with a small mean recovery error (MRE) of 3% (0.3%) under 1% H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> /air (1000 ppm NH <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> /air) gas at 25 °C.

Qualitative properties of the minimum sum-of-squares clustering problem

Fundamental qualitative properties of the minimum sum-of-squares clustering problem are established in this paper. We prove that the problem always has a global solution and, under a mild condition, the global solution set is finite. Moreover, the components of each global solution can be computed by an explicit formula. Based on a new concept of non-trivial local solution, we get necessary conditions for a system of centroids to be such a local solution. Interestingly, these necessary conditions are also sufficient ones. Finally, it is proved that the optimal value function is locally Lipschitz, the global solution map is locally upper Lipschitz, and the local solution map has the Aubin property, provided that the original data points are distinct. The obtained complete characterizations of the non-trivial local solutions allow one to understand better the performance of not only the k-means algorithm, but also of other solution methods for the problem in question.

Precision measurement and evaluation of flatness error for the aero-engine rotor connection surface based on convex hull theory and an improved PSO algorithm

The precision measurement and evaluation of the geometry form error for high-end precision components is one of the most important tasks for metrologists. In this paper, a novel hybrid flatness error evaluation method is proposed to measure and evaluate the flatness error for large-scale point cloud data. This method is based on the convex hull algorithm and the improved particle swarm optimization (IPSO) algorithm. Not only are a large number of redundant measurement data points effectively removed through the convex hull algorithm, but also this operation does not lose any valuable information for flatness error evaluation. Then an accurate mathematical model for flatness error evaluation is established based on the minimum zone criteria recommended by ISO, and then the IPSO algorithm is adopted to solve this complex non-linear optimization problem. In particular, the non-linear dynamic inertia weights w and learning factors c1, c2 are introduced to improve the calculation accuracy. Finally, the proposed method is verified through an experiment measuring and evaluating the flatness error of an aero-engine rotor connection surface. The effective measurement data points are reduced from 18 000 to 856, or 4.76%, which is significant compared to the original data points. The flatness error evaluation result is 369 µm according to the IPSO method that we propose. Last but not least, we compare our method with the traditional least squares method and unmodified particle swarm optimization algorithm. The calculation results indicate that the evaluation accuracy is increased by 21 µm and 44 µm, respectively.

CGD: Multi-View Clustering via Cross-View Graph Diffusion

Graph based multi-view clustering has been paid great attention by exploring the neighborhood relationship among data points from multiple views. Though achieving great success in various applications, we observe that most of previous methods learn a consensus graph by building certain data representation models, which at least bears the following drawbacks. First, their clustering performance highly depends on the data representation capability of the model. Second, solving these resultant optimization models usually results in high computational complexity. Third, there are often some hyper-parameters in these models need to tune for obtaining the optimal results. In this work, we propose a general, effective and parameter-free method with convergence guarantee to learn a unified graph for multi-view data clustering via cross-view graph diffusion (CGD), which is the first attempt to employ diffusion process for multi-view clustering. The proposed CGD takes the traditional predefined graph matrices of different views as input, and learns an improved graph for each single view via an iterative cross diffusion process by 1) capturing the underlying manifold geometry structure of original data points, and 2) leveraging the complementary information among multiple graphs. The final unified graph used for clustering is obtained by averaging the improved view associated graphs. Extensive experiments on several benchmark datasets are conducted to demonstrate the effectiveness of the proposed method in terms of seven clustering evaluation metrics.

Does Geographical Discrimination Exist in Online Lending in China: An Empirical Study Based on Chinese Loan Platform Renren

Background: Online lending has developed rapidly in China in recent years, into a typical Internet financial model. China’s online lending related issues have received widespread attention from scholars. Methods: This study used 396,634 order data-points (935,037 original order data-points) from the Renren Loan website since its inception in January 2017. We used ordinary least squares (OLS) regression to study the problem of geographical discrimination in online lending in China, and we conducted two robust tests. Results: Studies have shown that significant geographical discrimination exists in China’s online lending market. From the perspective of the lender, different investment intentions exist for borrowers from various regions, thereby leading to variations in the success rates of loans. From the perspective of the borrower, the belief exists that borrowers from different regions will have varying interest rates because of the effect of geographical discrimination. Conclusion: We believe that geographical discrimination is due to the effects of the economic, financial, educational, and ethnic conditions of the borrower’s location on willingness to invest and the success rate of borrowing. However, borrowers’ self-discrimination is primarily related to economic and ethnic differences among provinces.

Identification of critical process parameters of Jinqing alcohol precipitation of Reduning Injection by big data

Lonicerae Japonicae Flos and Artemisiae Annuae Herba(LA or Jinqing) alcohol precipitation has various process parameters and complex process mechanism, and is one of the key units for manufacturing Reduning Injection. In order to identify the critical process parameters(CPPs) affecting the weight of the extract produced from the alcohol precipitation process, 259 batches of historical production data from 2017 to 2018 were collected, with a total of 829 318 data points. These data showed characteristics of large data, such as a large data volume, a low value density, and diverse sources. The data cleaning and feature extraction were first performed, and 48 feature variables were selected. The original data points were reduced to 9 936. Then, a combination of Pearson correlation analysis and grey correlation analysis were used to screen out 15 potential critical process parameters(pCPPs). After that, the partial least squares(PLS) was used in prediction of the weight of the extract, proving that the performance of predictive model based on 15 pCMAs is equivalent to that of predictive model based on 48 feature variables. The variable importance in projection(VIP) index was used to identify 9 CPPs, including 2 alcohol precipitation supernatant volume parameters, 4 initial extract weight parameters and 3 added alcohol volume parameters. As a result, the number of data points was 1 863, accounting for 0.28% of the original data. The big data analysis approach from a holistic point of view can effectively increase the value density of the original data. The critical process parameters obtained can help to accurately describe the quality transfer mechanism of the Jinqing alcohol precipitation process.

A calibration data curve fitting method based on matrix orthogonal triangulation

This paper proposes a calibration data curve fitting method based on matrix orthogonal triangulation, which is applied in the field of ground calibration technology to solve the problem of low curve fitting precision and poor stability of the existing calibration data processing method. Firstly, the sampling data outputted by the infrared measuring system at different black body temperatures is decoded to obtain image data of a plurality of sampling points; the image gray mean value of each sampling point is calculated according to the image data of each sampling point, and the black body of each sampling point is obtained. The temperature value is converted into the black body radiation brightness value to obtain a plurality of original sampling point data; according to the linear original least squares principle, the matrix orthogonal triangulationmethod is used to solve the fit curve of the black body temperature-image gray value and black body radiance - the image gray value. The above method can fit straight lines and multiple curves, and has high fitting precision and good stability.

Calculation of Space NURBS Curve Interpolation Error

Aiming at the problem of low accuracy of interpolation error calculation of existing NURBS curves, an approximate method for the distance between a point and a NURBS interpolation curve is proposed while satisfying the accuracy of the solution. Firstly, the minimum parameter interval of the node vector corresponding to the data point under test in the original data point sequence is determined, and the parameter interval is subdivided according to the corresponding step size, and the corresponding parameter value is obtained. Secondly, the distance from the measured point to the NURBS curve is calculated, and the nearest distance is found out. The node interval is subdivided again on one side of the nearest distance. Finally, the distance between the data point to be measured and each subdivision point is calculated again, and the minimum distance is taken as the interpolation error between the point and the NURBS curve. The simulation results of actual tool position data show that this method can more accurately obtain the error of spatial NURBS interpolation curve.

Supervised Dimension Reduction by Local Neighborhood Optimization for Image Processing

Background: Subspace learning-based dimensionality reduction algorithms are important and have been popularly applied in data mining, pattern recognition and computer vision applications. They show the successful dimension reduction when data points are evenly distributed in the high-dimensional space. However, some may distort the local geometric structure of the original dataset and result in a poor low-dimensional embedding while data samples show an uneven distribution in the original space. Methods: In this paper, we propose a supervised dimension reduction method by local neighborhood optimization to disposal the uneven distribution of high-dimensional data. It extends the widely used Locally Linear Embedding (LLE) framework, namely LNOLLE. The method considers the class label of the data to optimize local neighborhood, which achieves better separability inter-class distance of the data in the low-dimensional space with the aim to abstain holding together the data samples of different classes while mapping an uneven distributed data. This effectively preserves the geometric topological structure of the original data points. Results: We use the presented LNOLLE method to the image classification and face recognition, which achieves a good classification result and higher face recognition accuracy compared with existing manifold learning methods including popular supervised algorithms. In addition, we consider the reconstruction of the method to solve noise suppression for seismic image. To the best of our knowledge, this is the first manifold learning approach to solve high-dimensional nonlinear seismic data for noise suppression. Conclusion: The experimental results on forward model and real seismic data show that LNOLLE improves signal to noise ratio of seismic image compared with the widely used Singular Value Decomposition (SVD) filtering method.

One center's experience developing a burn outpatient registry.

Recent advances in burn care have resulted in the transition of care from inpatient to outpatient. There is a growing appreciation that with improved survival, meaningful markers of quality need to include recovery of form, function, and reconstruction. Capture of the data describing care delivered in the outpatient setting is being missed. Development of our outpatient database included providers, registrar, program manager, and outpatient nursing staff. Data points were included if they described the population, and epidemiology of our patients, were useful for programmatic changes and improvements as well as anticipated research focus areas. The database platform chosen was Midas+™ because it was in use by hospital quality and integrated with the electronic medical record. Fields were customized based on changing program needs and are updated for new programs or outcomes measures. Reports can be easily built and both outpatients and inpatients are included. This allows for longitudinal tracking of burn patients. Ongoing additions to original data points include variables to track outcomes related to laser therapy for scar management, time to custom garment donning, and to track functional outcomes. Epidemiologic data collected is used to target high-risk populations for prevention and outreach efforts. Outcome data is used for evaluation of programs and care. High quality databases serve to measure effectiveness of care and offer insight for areas of improvement. There is a clear need for inclusion of outpatient activity in the National Burn Registry (NBR).

Application of a fully deep convolutional neural network to the automation of tooth segmentation on panoramic radiographs.

To evaluate a fully deep learning mask region-based convolutional neural network (R-CNN) method for automated tooth segmentation using individual annotation of panoramic radiographs. In total, 846 images with tooth annotations from 30 panoramic radiographs were used for training, and 20 panoramic images as the validation and test sets. An oral radiologist manually performed individual tooth annotation on the panoramic radiographs to generate the ground truth of each tooth structure. We used the augmentation technique to reduce overfitting and obtained 1024 training samples from 846 original data points. A fully deep learning method using the mask R-CNN model was implemented through a fine-tuning process to detect and localize the tooth structures. For performance evaluation, the F1 score, mean intersection over union (IoU), and visual analysis were utilized. The proposed method produced an F1 score of 0.875 (precision: 0.858, recall: 0.893) and a mean IoU of 0.877. A visual evaluation of the segmentation method showed a close resemblance to the ground truth. The method achieved high performance for automation of tooth segmentation on dental panoramic images. The proposed method might be applied in the first step of diagnosis automation and in forensic identification, which involves similar segmentation tasks.

CFD-based response surface methodology for rapid thermal simulation and optimal design of data centers

ABSTRACT In the design of large-scale data centres, CFD is used widely but very time-consuming with intensive computational resource requirement. When it comes to near real-time thermal control or optimizing multiple design parameters of data centres, this method becomes impractical. In this paper, response surface methodology (RSM) based on radial basis function (RBF) is used to significantly reduce the running time while maintaining a good accuracy. In the first application, by using 5%, 10%, and 20% of the original CFD data, the temperature profiles of the three corresponding cases are reconstructed based on RSM. Three reconstructed temperature profiles are then compared to the full temperature profile of a data centre model. The method shows good agreement with the CFD simulation result, especially for the case of 20% utilization of the original CFD data points. In the later application, RSM is used for generating a large set of generations during a two-objective optimization process which uses the genetic algorithm as its main engine. With three investigated design parameters including mass flow inlet, inlet temperature, and server heat load, the goal is to minimize both the temperature difference and the maximum temperature inside the data centre. The outcome shows a desirable range of design input parameters for a data centre. Highlights Response surface model is trained using high fidelity CFD simulation data Radial basis function is found to show superior advantages in constructing response surface Rapid thermal profile reconstruction for data centre using response surface method is illustrated CFD-based response surface method for data centre optimization process is investigated

Membership Affinity Lasso for Fuzzy Clustering

Fuzzy clustering generates a membership vector for each data point in the dataset to indicate its belongingness to different clusters. This procedure can be regarded as an encoding process and the obtained vectors of memberships are the new representations of original data. Naturally, the affinities between new representations or the vectors of memberships should be consistent with the ones between original data points. For example, the data points close to each other should also take similar membership vectors. Such constraints on the affinities of memberships are valuable prior knowledge that should be imposed to the objective function of fuzzy clustering for better performance. To this end, we introduce the membership affinity lasso for fuzzy clustering in this paper. Utilizing alternating direction method of multipliers, an efficient approach is derived to optimize the general membership affinity lasso regularized fuzzy clustering model in offline manner. As illustrative examples, three new fuzzy clustering algorithms with the membership affinity lasso are proposed. Experiments on the synthetic and real data demonstrate the superiority and flexibility of the proposed algorithms.