Cross-Project Software Defect Prediction Based on SMOTE and Deep Canonical Correlation Analysis

Cross-Project Defect Prediction(CPDP)is a method that utilizes historical data from other source projects to train predictive models for defect prediction in the target project.However,existing CPDP methods only consider linear correlations between features(indicators)of the source and target projects.These models are not capable of evaluating non-linear correlations between features when they exist,for example,when there are differences in data distributions between the source and target projects.As a result,the performance of such CPDP models is compromised.In this paper,this paper proposes a novel CPDP method based on Synthetic Minority Oversampling Technique(SMOTE)and Deep Canonical Correlation Analysis(DCCA),referred to as S-DCCA.Canonical Correlation Analysis(CCA)is employed to address the issue of non-linear correlations between features of the source and target projects.S-DCCA extends CCA by incorporating the MlpNet model for feature extraction from the dataset.The redundant features are then eliminated by maximizing the correlated feature subset using the CCA loss function.Finally,cross-project defect prediction is achieved through the application of the SMOTE data sampling technique.Area Under Curve(AUC)and F1 scores(F1)are used as evaluation metrics.This paper conducted experiments on 27 projects from four public datasets to validate the proposed method.The results demonstrate that,on average,our method outperforms all baseline approaches by at least 1.2%in AUC and 5.5%in F1 score.This indicates that the proposed method exhibits favorable performance characteristics.

Canonical Correlation Analysis of Agronomic Characters of Brassica juncea in Western China

[Objective] The study aimed at exploring the relationship among the agronomic characters of B. juncea in western China, in order to provide scientific basis for the breeding of B. juncea in western China. [Method] 39 B. juncea materials from western China were used for the canonical correlation analysis, and canonical correlations between each pair of the four ecological character （containing 18 variables） were verified, including yield characters （5 variables）, caulis characters （6 variables）, branch characters （3 variables） and pod characters （3 variables）. [Result] Yield per plant of B. juncea in western China suffered a tremendous influence from effective pod number per plant while was not significantly affected by the total pod number per plant, seed number per pod and 1 000-seed weight; the most important character related with the yield character of B. juncea in western China was caulis character, followed by the branch character and pod character; yield characters, caulis characters, branch characters and pod characters of B. juncea in western China were closely correlated. [Conclusion] In order to improve the yield characters of B. juncea in western China, caulis characters should be focused on, followed by branch characters and pod characters; rapeseed varieties with high performance in total pod number per plant and effective pod number per plant should be chosen through the perspectives of effective branch number, plant height, pod number of main inflorescence, fruit stalk number of main inflorescence and other traits, while rapeseed varieties with high performance in seed number per pod and 1 000-seed weight should be chosen through the perspectives of beak length and other traits.

Feature Fusion Multi-View Hashing Based on Random Kernel Canonical Correlation Analysis

Hashing technology has the advantages of reducing data storage and improving the efficiency of the learning system,making it more and more widely used in image retrieval.Multi-view data describes image information more comprehensively than traditional methods using a single-view.How to use hashing to combine multi-view data for image retrieval is still a challenge.In this paper,a multi-view fusion hashing method based on RKCCA(Random Kernel Canonical Correlation Analysis)is proposed.In order to describe image content more accurately,we use deep learning dense convolutional network feature DenseNet to construct multi-view by combining GIST feature or BoW_SIFT(Bag-of-Words model+SIFT feature)feature.This algorithm uses RKCCA method to fuse multi-view features to construct association features and apply them to image retrieval.The algorithm generates binary hash code with minimal distortion error by designing quantization regularization terms.A large number of experiments on benchmark datasets show that this method is superior to other multi-view hashing methods.

Study on soil water characteristics of tobacco fields based on canonical correlation analysis

In order to identify the principal factors influencing soil water characteristics （SWC） and evaluate SWC effectively, the multivariate-statistical canonical correlation analysis （CCA） method was used to study and analyze the correlation between SWC and soil physical and chemical properties. Twenty-two soil samples were taken from 11 main tobacco-growing areas in Guizhou Province in China and the soil water characteristic curves （SWCC） and basic physical and chemical properties of the soil samples were determined. The results show that： （1） The soil bulk density, soil total porosity and soil capillary porosity have significant effects on SWC of tobacco fiels. Bulk density and total porosity are positively correlated with soil water retention characteristics （SWRC）, and soil capillary porosity is positively correlated with soil water supply characteristics （SWSC）. （2） Soil samples from different soil layers at the same soil sampling point show similarity or consistency in SWC. Inadequate soil water supply capability and imbalance between SWRC and SWSC are problems of tobacco soil. （3） The SWC of loamy clay are generally superior to those of silty clay loam.

Novel DDoS Feature Representation Model Combining Deep Belief Network and Canonical Correlation Analysis

Distributed denial of service(DDoS)attacks launch more and more frequently and are more destructive.Feature representation as an important part of DDoS defense technology directly affects the efficiency of defense.Most DDoS feature extraction methods cannot fully utilize the information of the original data,resulting in the extracted features losing useful features.In this paper,a DDoS feature representation method based on deep belief network(DBN)is proposed.We quantify the original data by the size of the network flows,the distribution of IP addresses and ports,and the diversity of packet sizes of different protocols and train the DBN in an unsupervised manner by these quantified values.Two feedforward neural networks(FFNN)are initialized by the trained deep belief network,and one of the feedforward neural networks continues to be trained in a supervised manner.The canonical correlation analysis(CCA)method is used to fuse the features extracted by two feedforward neural networks per layer.Experiments show that compared with other methods,the proposed method can extract better features.

ASSESSMENT OF INFLUENCE IN CANONICAL CORRELATION ANALYSIS

By expanding the perturbation of covariance matrix in the powers of er-ror term,the influence functions for five canonical measurements in CCA are devel-oped and three sample versions are given.For generalized correlation coefficient p_z,the influence function is a quadratic form of r.v.z,and its distribution is considered.A practical example iUustrates the utility of the proposed influence functions.

SPATIAL REGULARIZATION OF CANONICAL CORRELATION ANALYSIS FOR LOW-RESOLUTION FACE RECOGNITION

Canonical correlation analysis ( CCA ) based methods for low-resolution ( LR ) face recognition involve face images with different resolutions ( or multi-resolutions ), i.e.LR and high-resolution ( HR ) .For single-resolution face recognition , researchers have shown that utilizing spatial information is beneficial to improving the recognition accuracy , mainly because the pixels of each face are not independent but spatially correlated.However , for a multi-resolution scenario , there are no related works.Therefore , a method named spatial regularization of canonical correlation analysis ( SRCCA ) is developed for LR face recognition to improve the performance of CCA by the regularization utilizing spatial information of different resolution faces.Furthermore , the impact of LR and HR spatial regularization terms on LR face recognition is analyzed through experiments.

Canonical correlation analysis to land-use structure and its driving forces——Taking Yulin Prefecture as an example

In this paper, one of the most classical statistical methods, Canonical Correlation Analysis (CCA) is applied to identify quantitatively the driving forces of landuse structure in Yulin Prefecture. The main analysis is carried out through the software SPSS with the data on the level of towns and townships in 1992. The results indicate that landuse structure is determined by comprehensive action of different factors. Landuse structure with rural characteristics is mainly determined by geographical factors such as the elevation, temperature and precipitation, while the landuse structure with urban characteristics is mainly determined by demographic and socioeconomic conditions. At the same time, tests were carried out through the canonical correlation coefficient and redundancy analysis.

ON USING NON-LINEAR CANONICAL CORRELATION ANALYSIS FOR VOICE CONVERSION BASED ON GAUSSIAN MIXTURE MODEL

Voice conversion algorithm aims to provide high level of similarity to the target voice with an acceptable level of quality.The main object of this paper was to build a nonlinear relationship between the parameters for the acoustical features of source and target speaker using Non-Linear Canonical Correlation Analysis(NLCCA) based on jointed Gaussian mixture model.Speaker indi-viduality transformation was achieved mainly by altering vocal tract characteristics represented by Line Spectral Frequencies(LSF).To obtain the transformed speech which sounded more like the target voices,prosody modification is involved through residual prediction.Both objective and subjective evaluations were conducted.The experimental results demonstrated that our proposed algorithm was effective and outperformed the conventional conversion method utilized by the Minimum Mean Square Error(MMSE) estimation.

Canonical Correlation Analysis and climate research

Correlation analysis as used by meteorologists and oceanographers is a tool for the analysisof the spacial or temporal variability of physical fields. In his notes, Dr. Hasselmann pro-posed to combine correlation analysis and linear regression analysis in climate prediction re-search. The main idea is to decompose the physical field into its principal oscillation patterns.

A NOVEL ALGORITHM FOR VOICE CONVERSION USING CANONICAL CORRELATION ANALYSIS

A novel algorithm for voice conversion is proposed in this paper. The mapping function of spectral vectors of the source and target speakers is calculated by the Canonical Correlation Analysis (CCA) estimation based on Gaussian mixture models. Since the spectral envelope feature remains a majority of second order statistical information contained in speech after Linear Prediction Coding (LPC) analysis, the CCA method is more suitable for spectral conversion than Minimum Mean Square Error (MMSE) because CCA explicitly considers the variance of each component of the spectral vectors during conversion procedure. Both objective evaluations and subjective listening tests are conducted. The experimental results demonstrate that the proposed scheme can achieve better per- formance than the previous method which uses MMSE estimation criterion.

Multiple moving sources passive location based on multiset canonical correlation analysis

To solve the problem of multiple moving sources passive location, a novel blind source separa- tion （BSS） algorithm based on the muhiset canonical correlation analysis （MCCA） is presented by exploiting the different temporal structure of uncorrelated source signals first, and then on the basis of this algorithm, a novel multiple moving sources passive location method is proposed using time difference of arrival （TDOA） and frequency difference of arrival （FDOA） measurements. The key technique of this location method is TDOA and FDOA joint estimation, which is based on BSS. By blindly separating mixed signals from multiple moving sources, the multiple sources location problem can be translated to each source location in turn, and the effect of interference and noise can also he removed. The simulation results illustrate that the performance of the MCCA algorithm is very good with relatively light computation burden, and the location algorithm is relatively simple and effective.

Trace Lasso Regularization for Adaptive Sparse Canonical Correlation Analysis via Manifold Optimization Approach

Canonical correlation analysis(CCA)describes the relationship between two sets of variables by finding a linear combination that maximizes the correlation coefficient.However,in high-dimensional settings where the number of variables exceeds sample size,or in the case that the variables are highly correlated,the traditional CCA is no longer appropriate.In this paper,a new matrix regularization is introduced,which is an extension of the trace Lasso in the vector case.Then we propose an adaptive sparse version of CCA(ASCCA)to overcome these disadvantages by utilizing the trace Lasso regularization.The adaptability of ASCCA is that the sparsity regularization of canonical vectors depends on the sample data,which is more realistic in practical applications.The ASCCA model is further reformulated to an optimization problem on the Riemannian manifold.Then we adopt a manifold inexact augmented Lagrangian method to solve the resulting optimization problem.The performance of the ASCCA model is compared with some existing sparse CCA techniques in different simulation settings and real datasets.

Asymptotic distributions in the projection pursuit based canonical correlation analysis

In this paper, associations between two sets of random variables based on the projection pursuit (PP) method are studied. The asymptotic normal distributions of estimators of the PP based canonical correlations and weighting vectors are derived.

Convergence rate of kernel canonical correlation analysis

Kernel canonical correlation analysis(CCA) is a nonlinear extension of CCA,which aims at extract-ing the information shared by two random variables. It has wide applications in many fields,such as information retrieval. This paper gives the convergence rate analysis of kernel CCA under some approximation conditions and some suggestions on how to choose the regularization parameter. The result shows that the convergence rate only depends on two parameters:the rate of regularization parameter and the decay rate of eigenvalues of compact operator VY X,and it gives better understanding of kernel CCA.

Multi-view dimensionality reduction via canonical random correlation analysis

Canonical correlation analysis （CCA） is one of the most well-known methods to extract features from multi- view data and has attracted much attention in recent years. However, classical CCA is unsupervised and does not take discriminant information into account. In this paper, we add discriminant information into CCA by using random cross- view correlations between within-class samples and propose a new method for multi-view dimensionality reduction called canonical random correlation analysis （RCA）. In RCA, two approaches for randomly generating cross-view correlation samples are developed on the basis of bootstrap technique. Furthermore, kernel RCA （KRCA） is proposed to extract nonlinear correlations between different views. Experiments on several multi-view data sets show the effectiveness of the proposed methods.

SSTA SIGNAL CHARACTERISTIC ANALYSIS OVER THE INDIAN OCEAN DURING RAINY SEASON IN CHINA

The teleconnection distribution characteristics of sea surface temperature (SST) over the India Ocean and the precipitation during rainy season in China were studied by using the methods of EOF and CCA. The results indicate that the change of SST field will affect the change of rain belt during rainy seasons in China, and greatly affect the precipitation in northwest and southwest China, the Yangzi and Yellow River downstream basins. Strong signal phenomena of SSTA over India Ocean were revealed that showed the anoma-lous distribution of drought and flood in China. It shows that the precipitation during rainy seasons in China may be forecast by analyzing SST distribution characteristics over the India Ocean.

Canonical correlation analysis of hydrological response and soil erosion under moving rainfall

The impacts of rainfall direction on the degree of hydrological response to rainfall properties were investigated using comparative rainfall-runoff experiments on a small-scale slope(4 m×1 m),as well as canonical correlation analysis(CCA).The results of the CCA,based on the observed data showed that,under conditions of both upstream and downstream rainfall movements,the hydrological process can be divided into instantaneous and cumulative responses,for which the driving forces are rainfall intensity and total rainfall,and coupling with splash erosion and wash erosion,respectively.The response of peak runoff(Pr)to intensity-dominated rainfall action appeared to be the most significant,and also runoff(R)to rainfall-dominated action,both for upstream-and downstream-moving conditions.Furthermore,the responses of sediment erosion in downstream-moving condition were more significant than those in upstream-moving condition.This study indicated that a CCA between rainfall and hydrological characteristics is effective for further exploring the rainfall-runoff-erosion mechanism under conditions of moving rainfall,especially for the downstream movement condition.

Regularized canonical correlation analysis with unlabeled data

In standard canonical correlation analysis (CCA), the data from definite datasets are used to estimate their canonical correlation. In real applications, for example in bilingual text retrieval, it may have a great portion of data that we do not know which set it belongs to. This part of data is called unlabeled data, while the rest from definite datasets is called labeled data. We propose a novel method called regularized canonical correlation analysis (RCCA), which makes use of both labeled and unlabeled samples. Specifically, we learn to approximate canonical correlation as if all data were labeled. Then, we describe a generalization of RCCA for the multi-set situation. Experiments on four real world datasets, Yeast, Cloud, Iris, and Haberman, demonstrate that, by incorporating the unlabeled data points, the accuracy of correlation coefficients can be improved by over 30%.

Analysis on the Influencing Factors of Job Burnout of Nurses in Haikou Tertiary Hospital

Objective: To understand the influencing factors of job burnout among nurses in Haikou 3A hospital and explore its direct and indirect effects, so as to provide a scientific basis for the work efficiency of nursing staff. Methods: Between November 2, 2015 and November 2015, using multi stage random sampling, self-administered questionnaire survey was conducted among 1049 nursing staff, using the path analysis method to study the effect of direct and indirect factors effect. Results: The total score of job burnout of nurses was 38.44 ± 7.55, high occupational burnout was 0.9%, moderate occupational burnout was 66.5%, and low occupational burnout was 32.6%. The scores of job burnout were compared among the nurses with different titles, and less achievement (F = 8.342, P < 0.001) and depersonalization (F = 3.12, P = 0.025) were statistically significant. Nurses’ Job Burnout and job stressors were the first, and the canonical correlation coefficient was 0.4397 (F = 20.54, P < 0.0001), indicating that the more problems existed in patient care, the greater the degree of emotional exhaustion. The first canonical correlation coefficient of job burnout and job satisfaction of nurses was 0.3791 (F = 12.8, P < 0.0001), indicating that the better the family and work balance, the less individualized nurses were. The path analysis results showed that the 4 dimensions of job stressors (management and interpersonal problems) is positive, the direct effect of the strongest (0.219), the total effect of sort of work pressure source of 4 dimensions (0.245) > 5 dimensions of work pressure source (0.125) > title (−0.112) job satisfaction scores (−0.097). Conclusion: Job stress, job satisfaction and job title are the factors that affect job burnout. The 4 and the direct and indirect effects of job stressors are the strongest, and measures should be taken to solve these problems.