Nonlinear Principal Component Research Articles

Improving corporate bond recovery rate prediction using multi-factor support vector regressions

In the multi-factor framework described in this paper, we use instrument-specific characteristics, several macroeconomic variables, and industry-specific characteristics as our explanatory variables for predicting recovery rates for corporate bonds. By including the principal components derived from a large number of macroeconomic variables, all three least-squares support vector regression methods, as well as the ordinary linear regression, exhibit higher out-of-sample predictive accuracy than the models that included only the few macroeconomic variables suggested in the literature. We compare the prediction accuracies of all techniques by incorporating sparse principal components, nonlinear principal components from an auto-associative neural network, and kernel principal components. Our results show that sparse principal components generate more interpretable and accurate estimations compared to the other principal component techniques. Moreover, we apply gradient boosting to generate a ranking of the 104 macroeconomic variables, from best to worst, based on their prediction power in recovery rate estimation. The three categories with the most informative macroeconomic predictors are micro-level factors, business cycle variables, and stock market indicators.

Multi-objective ensemble forecasting with an application to power transformers

In this paper we present an ensemble time series forecasting algorithm using evolutionary multi-objective optimization algorithms to predict dissolved gas contents in power transformers. In this method, the correlation between each individual dissolved gas and other transformers’ features such as temperature characteristics and loading history is first determined. Then, a non-linear principal component analysis (NLPCA) technique is applied to extract the most effective time series from the highly correlated features. Afterwards, the forecasting algorithms are trained using a cross validation technique. In addition, evolutionary multi-objective optimization algorithms are used to select the most accurate and diverse group of forecasting algorithms to construct an ensemble. Finally, the selected ensemble is examined to predict the value of the dissolved gases on the testing set. The results of one day, two day, three day, and four day ahead forecasting are presented which show higher accuracy and reliability of the proposed method compared with other statistical methods.

TECHNOLOGIES OF THE FOURTH INDUSTRIAL REVOLUTION: RESEARCH OF TECHNOLOGICAL ACCEPTANCE OF INDUSTRIAL COMPANIES WITH THE HELP OF NONLINEAR PRINCIPAL COMPONENT ANALYSIS METHOD (NLPCA)

Industrial companies’ readiness to accept the industry technologies 4.0 is studied and results of industrial companies’ questioning are analyzed in the work. The nonlinear principal component analysis method (NLPCA) is applied as a data processing instrument in understanding the Gifi system taking into account the variegated statistic nature of the questioning indexes.

Interview

Dr Xueyun Wei from Jiangsu University of Science and Technology, China, talks to Electronics Letters about the work behind the paper ‘Application of kernel PCA for foetal electrocardiogram estimation’, page 340. Dr Xueyun Wei I joined a team of scientific researchers who, using both advanced signal processing and detection technology, focus on biomedical signal processing and methods that can help to diagnose diseases. At the moment, my research focuses on the non-invasive detection of foetal electrocardiogram (ECG). Foetal ECG monitoring using abdominal ECG signals involves the detection of signals with low amplitudes that are produced by the foetal heartbeat. These signals are often overwhelmed by large interference and noise. The maternal ECG is the predominant source of interference, which is the most important consideration that must be accounted for when estimating foetal ECG signals. Various techniques have been employed to address foetal ECG estimation, including correlation techniques, adaptive filtering, wavelet analysis, blind source separation, neural networks and singular value decomposition. When employing an analysis method, such as principal component analysis, the powerful maternal ECG is considered to be the principal component as it dominates the foetal ECG and other background noises. Utilising the correlation between the maternal components in different lead ECG signals, allows maternal components from abdominal signals to be removed and foetal ECG to be estimated. In our Letter, we have reported a new method to estimate the antepartum foetal ECG signal using kernel principal component analysis (kPCA). We show that this method can reveal nonlinearities in the principal components of signals measured with different leads, thereby allowing the maternal components in different lead signals to be extracted. When compared to the popular independent component analysis method, experimental data from the proposed method has shown strong nonlinear analysis ability for foetal ECG estimation. The work reported in our Letter shows strong nonlinearity between the maternal components of every collected signal due to observed diversity in the signal propagation path. Kernel PCA can be considered to be a non-linear form of PCA that can extract the non-linear principal component from multi-dimensional data. Thus, this method can be applied to ECG signals that have been recorded using multiple leads to eliminate the maternal ECG components and accurately estimate foetal ECG signals. Selecting an appropriate kernel function and parameters based on multi-dimensional nonlinear data is the only challenge associated with this method. Although many signal processing methods have been used to extract foetal ECG information from the abdominal ECG signal, these methods still experience problems that limit their clinical application, such as low extraction accuracy and reliability difficulties. Most of the existing research methods use the linear mixed model or the local linear mixed model, which lack the ability to perform extensive nonlinear analysis. This work provides a new nonlinear un-mixing model for foetal ECG extraction, which could be used for clinical applications in the near future. We are planning to apply this method in appropriate technology as soon as possible. Changes in the foetal heart function are due to the regulating function of the central nervous system. Therefore, foetal heart monitoring plays a very important role in foetal disease/deficiency diagnosis. I am also studying big data issues associated with biomedical signals. At the moment, data storage capacities and modern technology has exceeded the intellectual ability of people. This issue is particularly prevalent for contemporary medical service data, which is ever-increasing. However, traditional statistical techniques and data managerial tools cannot meet current demands. The demand for quantitative diagnosis and precise treatment is increasing in modern medicine. Therefore, the application of biomedical signal processing is a field that is rapidly expanding. This field is full of challenging and innovative opportunities. The excitement it brings to researchers is like climbing Mount Qomolangma (Everest), making people full of passion and expectation.

A new soil quality index based on morpho-pedological indicators as a site-specific web service applied to olive groves in the Province of Jaen (South Spain)

Soil quality has become a fundamental concept in soil science and agriculture, but it can be difficult to apply its theoretical and experimental approaches to poorly surveyed zones where precision techniques are far from being applied. In this paper, we propose a new technique that enables little-used qualitative morpho-pedological data to be managed and integrated into a single Field Soil Quality Index (FSQI). Nonlinear Principal Component Analysis (NLPCA), a technique able to handle categorical data, is applied here to deal with morpho-pedological indicators. When categorical values are transformed, they can be properly analyzed and interpreted. This procedure requires less expert knowledge, so it can help soil quality assessments by non-experts. We applied the FSQI protocol to soils in the most important olive-growing area in the world, Jaen Province (Southern Spain), which has serious problems with soil degradation and erosion. First, a soil database for the study area was compiled, including 18 morphological attributes for 131 surface horizons belonging to eight Land Use Types. Secondly, the NLPCA provides optimal scalings and attribute weights that transform and integrate morphological indicators into a simple weighted additive index (FSQI). Thirdly, the scaling functions and weights found were applied to the same attributes of an evaluation set comparing two soil management types (conventional vs. organic) in olive groves. The FSQI means for the first (conventional) were significantly lower than in the organic groves (0.278 vs. 0.463, P < .05), which supports the validity of the index. A site-specific FSQI web service has been created to assist in decision-making in the study area, whose methodology can be generalized to other zones and crops.

Statistical assessment of rock burst potential and contributions of considered predictor variables in the task

Rock burst phenomenon is one of the major concerns in underground mining and excavating. Despite the significant progress in the field of phenomenon prediction, including data-driven researches, the need for developing reliable models is still felt. In this research, a database consisting of 188 distinct case histories was considered. Each case history contains some of the predictor variables “overburden thickness, maximum tangential stress in the boundary of opening, uniaxial compressive strength of rock, tensile strength of rock, stress ratio, brittleness ratio and elastic strain energy index” and one of the four defined classes (none or not-occurred, weak, moderate and strong) for the qualitative dependent variable “rock burst intensity”. After data preprocessing procedure (including outlier detection and substitution, factor analysis and nonlinear principal component analysis in which the two later did not lead to extract reasonable predictor components), 48 regression models were fitted for different original predictor variable arrangements, fitting datasets and link functions to obtain proper statistical equations for rock burst occurrence and intensity prediction. The best models, recognized based on fulfilling certain criteria and maximizing specific objectives, have proper classification accuracies, especially in comparison to those of the empirical criteria developed for rock burst prediction. Due to the presence pattern of the predictor variables in regression equations, a stepwise statistical strategy, consisting of normality condition evaluation, evaluation of mean equality and distribution function equality hypotheses and correlation analysis for predictor variables, was considered. The strategy led to conclude that the predictor variables “overburden thickness, tensile strength of rock and brittleness ratio” with the available case histories have insignificant contributions to predict rock burst phenomenon. By contrast, the predictor variables “maximum tangential stress, stress ratio, elastic strain energy index and uniaxial compressive strength of rock” have considerable contributions, decreasing in an ascending order. The application of non-dominated sorting genetic algorithm and multi-layer perceptron neural networks for solving the feature (i.e. variable) selection problem had a perfect adaption with the inferences.

Factors affecting energy-saving behaviours and energy efficiency investments in British households

The objective of this paper is to identify the socio-demographic, dwelling, and environmental factors that have the strongest influence on the daily energy-saving behaviours, the adoption of energy efficient appliances and the energy efficient retrofit investments made by British households. This study uses British household data from the “Survey of Public Attitudes and Behaviours towards the Environment” collected in 2009, and employs nonlinear principal components analysis (NLPCA), ordinary least squares (OLS) regression, and probit models. The results show different household profiles with specific features driving daily energy-saving behaviours and energy efficiency investments. Environmental variables are a good predictor of both energy-saving behaviours and investment in energy efficient appliances but not of energy efficient retrofit measures. Results of income and dwelling type variables with regard to energy-saving behaviours and energy efficient retrofit investments significantly diverge; in addition, interesting patterns emerge with respect to the respondents’ age, sex, and marital status. By evaluating and understanding the household and dwelling characteristics that affect energy-saving behaviours and energy efficiency investments, it is possible to obtain a clearer idea of where and how energy and emissions savings can be made, and to propose effective and targeted policies that promote energy-responsible lifestyles.

New Methods of Fracture Effect Forecast of CBM Wells Based on KPCA and Artificial Neural Networks

Based on the problems that coal bed methane (CBM) wells had no or low productions after fracturing, this paper combine the knowledge of logging technology, hydraulic fracturing technology, dewatering technology, optimization theory and artificial neural networks(ANN), a method of fracture effect forecast of CBM wells based on Kernel function of nonlinear principal component analysis (KPCA) and ANN is presented. This paper introduces the KPCA, ANN algorithm and sums up the influence of CBM well fracturing effect of the main factors. Through the method to the input of the model parameters were analyzed and the main parameters of the extraction, and based on this, established the ANN prediction model, and presents a case analysis. Application shows that the KPCA method combined with BP artificial neural network prediction of CBM well fracturing effect, simplified network structure, improve the computational speed, good practicability and reliability, is a worthy of promotion methods.

Spectral transformation based on nonlinear principal component analysis for dimensionality reduction of hyperspectral images

ABSTRACTManaging transmission and storage of hyperspectral (HS) images can be extremely difficult. Thus, the dimensionality reduction of HS data becomes necessary. Among several dimensionality reduction techniques, transform-based have found to be effective for HS data. While spatial transformation techniques provide good compression rates, the choice of the spectral decorrelation approaches can have strong impact on the quality of the compressed image. Since HS images are highly correlated within each spectral band and in particular across neighboring bands, the choice of a decorrelation method allowing to retain as much information content as possible is desirable. From this point of view, several methods based on PCA and Wavelet have been presented in the literature. In this paper, we propose the use of NLPCA transform as a method to reduce the spectral dimensionality of HS data. NLPCA represents in a lower dimensional space the same information content with less features than PCA. In these terms, aim of this research is focused on the analysis of the results obtained through the spectral decorrelation phase rather than taking advantage of both spectral and spatial compression. Experimental results assessing the advantage of NLPCA with respect to standard PCA are presented on four different HS datasets.

Assessing Effects of Temperature Change on Four Limonium Species from Threatened Mediterranean Salt-Affected Habitats

Evaluation of tolerance to temperatures in the germination stage is an important tool to foresee the possible effect of global warming. Moreover, establishing germination protocols for endemic and threatened species is most interesting for restoration and management plans that focus on habitat conservation. Seed germination was analysed in four species of the genus Limonium (L. virgatum, L. narbonense, L. girardianum and L. santapolense) growing in two saline habitats on the eastern coast of the Iberian Peninsula. The study aim was to compare responses of seed germination to different temperature regimes. Seeds were collected in two protected areas, ‘L´ Albufera de Valencia’ (province of Valencia) and ‘Clot de Galvany’ (province of Alicante). Responses to temperature were checked within a constant range, 10-30 °C at 5 °C intervals, by considering the environmental conditions at both sampling sites. The final germination percentage and mean germination time (MGT) were calculated after 30 days. In order to compare thermal responses, base temperature (Tb) and thermal time (S) were determined for each species. The possible correlations of soil parameters with the species´ germination pattern were also analysed. To identify distinct behavioural groups, a non-linear principal component analysis was performed. Significant differences between species were found in the velocity of germination. A fast response as opportunistic germination was observed in all the species. The base temperature and thermal time showed significant differences in competitiveness between species.

Examination of the structure of the traditional/secular-rational and survival/self-expression value dimensions in the model of Ronald Inglehart: The case of Serbia

The paper examines applicability of Inglehart?s model of dimensions of traditional/secular-rational values and survival/self expresion values, based on European Values Study data, conducted in Serbia in 2008. Dimensions of values are extracted by applying nonlinear principal component analysis, with asigning ordinal level of analysis. We present three-dimensional solution which separates dimension of survival/self expresion values into two dimensions. The deviations from the two-dimensional model were interpreted in accordance with the theory of revised modernization and thesis about value change in dependance of sociocultural paths of developement of society.

Distance metric learning-based kernel gram matrix learning for pattern analysis tasks in kernel feature space

Approaches to distance metric learning (DML) for Mahalanobis distance metric involve estimating a parametric matrix that is associated with a linear transformation. For complex pattern analysis tasks, it is necessary to consider the approaches to DML that involve estimating a parametric matrix that is associated with a nonlinear transformation. One such approach involves performing the DML of Mahalanobis distance in the feature space of a Mercer kernel. In this approach, the problem of estimation of a parametric matrix of Mahalanobis distance is formulated as a problem of learning an optimal kernel gram matrix from the kernel gram matrix of a base kernel by minimizing the logdet divergence between the kernel gram matrices. We propose to use the optimal kernel gram matrices learnt from the kernel gram matrix of the base kernels in pattern analysis tasks such as clustering, multi-class pattern classification and nonlinear principal component analysis. We consider the commonly used kernels such as linear kernel, polynomial kernel, radial basis function kernel and exponential kernel as well as hyper-ellipsoidal kernels as the base kernels for optimal kernel learning. We study the performance of the DML-based class-specific kernels for multi-class pattern classification using support vector machines. Results of our experimental studies on benchmark datasets demonstrate the effectiveness of the DML-based kernels for different pattern analysis tasks.

A new reconstruction-based auto-associative neural network for fault diagnosis in nonlinear systems

Auto-associative neural network (AANN) is a typical nonlinear principal component analysis method, which is widely used in industry for fault diagnosis purposes, especially in nonlinear systems. However, the basic AANN often suffers from “smearing effects” problems that may lead to misdiagnosis, particularly with regards to the complex faults involving multiple variables. In this work, a new reconstruction-based AANN (RBAANN) method is proposed to enhance the capacity of fault diagnosis. In RBAANN, a generic derivative equation is developed to investigate the effects of AANN model inputs on the prediction error between model inputs and outputs. Based on the derivative equation, the reconstruction-based index for single or multiple variables, which is defined as the minimum prediction error, is obtained by tuning the corresponding model inputs iteratively. However, without the prior knowledge of the real faulty variables, all the possible variable sets need to be evaluated by the reconstruction-based index, and this may result in an exhaustive search and cause a huge computational burden. Thus, a branch and bound algorithm is introduced into RBAANN to solve the variable selection problem. Finally, an efficient fault diagnosis strategy by integrating RBAANN and branch and bound algorithm (BAB-RBAANN) is implemented to further pinpoint the source of the detected faults. This BAB-RBAANN method can handle both single and multiple variable(s) faults for nonlinear systems without prior knowledge efficiently. The effectiveness of the proposed methods is evaluated on a validation example and an industrial example. Comparisons with other methods, including principal component analysis techniques, are also presented.

Damage detection methodology under variable load conditions based on strain field pattern recognition using FBGs, nonlinear principal component analysis, and clustering techniques

Structural health monitoring consists of using sensors integrated within structures together with algorithms to perform load monitoring, damage detection, damage location, damage size and severity, and prognosis. One possibility is to use strain sensors to infer structural integrity by comparing patterns in the strain field between the pristine and damaged conditions. In previous works, the authors have demonstrated that it is possible to detect small defects based on strain field pattern recognition by using robust machine learning techniques. They have focused on methodologies based on principal component analysis (PCA) and on the development of several unfolding and standardization techniques, which allow dealing with multiple load conditions. However, before a real implementation of this approach in engineering structures, changes in the strain field due to conditions different from damage occurrence need to be isolated. Since load conditions may vary in most engineering structures and promote significant changes in the strain field, it is necessary to implement novel techniques for uncoupling such changes from those produced by damage occurrence. A damage detection methodology based on optimal baseline selection (OBS) by means of clustering techniques is presented. The methodology includes the use of hierarchical nonlinear PCA as a nonlinear modeling technique in conjunction with Q and nonlinear-T2 damage indices. The methodology is experimentally validated using strain measurements obtained by 32 fiber Bragg grating sensors bonded to an aluminum beam under dynamic bending loads and simultaneously submitted to variations in its pitch angle. The results demonstrated the capability of the methodology for clustering data according to 13 different load conditions (pitch angles), performing the OBS and detecting six different damages induced in a cumulative way. The proposed methodology showed a true positive rate of 100% and a false positive rate of 1.28% for a 99% of confidence.

An improved extreme learning machine integrated with nonlinear principal components and its application to modeling complex chemical processes

In order to enhance the performance of extreme learning machine (ELM) in modeling complex chemical processes, an improved ELM integrated with nonlinear principal components is proposed. Firstly, an improved ELM (IELM) model is presented. The IELM has a special structure with two independent input subnets: a positive correlation subnet and a negative correlation subnet. The two independent input subnets are developed based on the correlation coefficient between input attributes and output attributes. The nonlinear principal components of original input attributes are extracted using input training neural network (ITNN). The extracted nonlinear principal components are connected to output layer nodes. Thus, the output nodes not only connect with the positive correlation subnet and the negative correlation subnet, but also with the extracted nonlinear principal components. Thus, an IELM integrated with nonlinear principal components (NPCs-IELM) model can be built. The effectiveness of the proposed NPCs-IELM is verified by modeling a high density polyethylene process. Simulation results indicate that the proposed NPCs-IELM can achieve higher accuracy and better stability.

A hybrid classifier based on nonlinear-PCA and deep belief networks with applications in dysphagia diagnosis

Traditional dysphagia prescreening diagnostic methods require doctors specialists to give patients a total score based on a water swallow test scale. This method is limited by the high dimensionality of the diagnostic elements in the water swallow test scale with heavy workload (Towards each patient, the scale requires the doctors give score for 18 diagnostic elements respectively) as well as the difficulties of extracting and using the diagnostic scale data’s non-linear features and hidden expertise information (Even with the scale scores, specific diagnostic conclusions are still given by expert doctors under the expertise). In this paper, a hybrid classifier model based on Nonlinear-Principal Component Analysis (NPCA) and Deep Belief Networks (DBN) is proposed in order to effectively extract the diagnostic scale data’s nonlinear features and hidden information and to provide the key scale elements’ locating methods towards the diagnostic results (The key scale elements that affect different diagnostic conclusions are given to improve the efficiency and pertinence of diagnosis and reduce the workload of diagnosis). We demonstrate the effectiveness of the proposed method using the frame of ‘information entropy theory’. Real dysphagia diagnosis examples from the China-Japanese Friendship Hospital are used to demonstrate applications of the proposed methods. The examples show satisfactory results compared to the traditional classifier.

Identifying Global Value Chain Governance Determinants in Kenya’s Horticultural Oriented Export Sector by Nonlinear Principal Component Analysis

The objective of this work is to empirically assess the determinants used in global value chain governance trade through interfirm relationship and their effect on competitive advantage of horticultural value chains in developing economies. The paper is based on a Kenyan survey of 83 horticultural exporters of fresh fruits and vegetables accessed out of 120 exporters registered with Fresh Produce Association of Kenya (FPEAK) and Horticultural Crops Directorate (HCD). It employs a nonlinear principal component analysis procedure of categorical principal component analysis (CATPCA). From the CATPCA output, five components considered to be key food value chain governance determinants affecting the export oriented horticultural value chain in Kenya were extracted and named as standards & certification, nature of transactions, level of chain integration, nature of contract and relational characteristics. The current research shows that with prioritisation of requisite food safety standards & certification schemes and discernment of transactional demands of heterogeneous export markets, horticultural exporters may expect to be competitive by means of improved product differentiation, process flows, logistical advantage and promotion of marketing & organisation management innovativeness. The implication therefore to the chain managers that standards & certification schemes are taking a critical position in determining food value chain governance especially through remote governance or governance from a distance other than the governance structure continuum of markets and hierarchies previously defined by transactional cost economics.

Inherent vulnerability assessment of rural households based on socio-economic indicators using categorical principal component analysis: A case study of Kimsar region, Uttarakhand

The recent trend of shifting focus from hazard centric drivers of vulnerability towards the social and economic drivers of vulnerability has led to a number of conceptual frameworks for social vulnerability assessment. Contributing towards this growing trend of social vulnerability assessment, this study proposes a framework to measure inherent vulnerability, which is centered on hazard generic and livelihood oriented socioeconomic factors of vulnerability. Inherent vulnerability is defined as the predisposition of a household to suffer harm. The study focuses on the mountainous communities in Kimsar region, located in Uttarakhand state, India. The communities in the region suffer from multiple stressors including extreme precipitation, drought, landslides, cloudbursts and flash floods. Vulnerability indicators with mixed scaling are used, to capture household’s perception and other socio-economic attributes, which contribute towards its inherent vulnerability. Data was collected by conducting household surveys in nine villages of Kimsar region. In order to process the indicators with mixed scaling, and obtain an empirical summary of the data set, the method of Non-Linear Principal Component Analysis was used for computing a household level Inherent Vulnerability Index. Results obtained revealed that principal components explaining a major variance in the data set were — access to employment opportunities, effectiveness of local government, access to food, occupational diversity, access to resources, educational attainment and access to water. It was observed that the villages of Dharkot, Kandakhal and Bhumiyakisar have the highest percentage of households, which were relatively less vulnerable to environmental stressors. Higher vulnerability was observed in majority of households in the village Kimsar, Ramjeewala and Malla Banas. A majority of households in Talla Banas, Jogiyana and Kasan were moderately vulnerable. Inherent vulnerability assessment has the potential to predict the future harm a household might suffer due to hazard events.

Shallow and deep learning for image classification

The paper is focused on the idea to demonstrate the advantages of deep learning approaches over ordinary shallow neural network on their comparative applications to image classifying from such popular benchmark databases as FERET and MNIST. An autoassociative neural network is used as a standalone program realized the nonlinear principal component analysis for prior extracting the most informative features of input data for neural networks to be compared further as classifiers. A special study of the optimal choice of activation function and the normalization transformation of input data allows to improve efficiency of the autoassociative program. One more study devoted to denoising properties of this program demonstrates its high efficiency even on noisy data. Three types of neural networks are compared: feed-forward neural net with one hidden layer, deep network with several hidden layers and deep belief network with several pretraining layers realized restricted Boltzmann machine. The number of hidden layer and the number of hidden neurons in them were chosen by cross-validation procedure to keep balance between number of layers and hidden neurons and classification efficiency. Results of our comparative study demonstrate the undoubted advantage of deep networks, as well as denoising power of autoencoders. In our work we use both multiprocessor graphic card and cloud services to speed up our calculations. The paper is oriented to specialists in concrete fields of scientific or experimental applications, who have already some knowledge about artificial neural networks, probability theory and numerical methods.

New tools for the visualization of biological pathways.

This paper presents several geometrically motivated techniques for the visualization of high-dimensional biological data sets. The Grassmann manifold provides a robust framework for measuring data similarity in a subspace context. Sparse radial basis function classification as a visualization technique leverages recent advances in radial basis function learning via convex optimization. In the spirit of deep belief networks, supervised centroid-encoding is proposed as a way to exploit class label information. These methods are compared to linear and nonlinear principal component analysis (autoencoders) in the context of data visualization; these approaches may perform poorly for visualization when the variance of the data is spread across more than three dimensions. In contrast, the proposed methods are shown to capture significant data structure in two or three dimensions, even when the information in the data lives in higher dimensional subspaces. To illustrate these ideas, the visualization techniques are applied to gene expression data sets that capture the host immune system's response to infection by the Ebola virus in non-human primate and collaborative cross mouse.