Nonlinear Data Fusion Research Articles

A novel method for data fusion over entity-relation graphs and its application to protein-protein interaction prediction.

Modern bioinformatics is facing increasingly complex problems to solve, and we are indeed rapidly approaching an era in which the ability to seamlessly integrate heterogeneous sources of information will be crucial for the scientific progress. Here, we present a novel non-linear data fusion framework that generalizes the conventional matrix factorization paradigm allowing inference over arbitrary entity-relation graphs, and we applied it to the prediction of protein-protein interactions (PPIs). Improving our knowledge of PPI networks at the proteome scale is indeed crucial to understand protein function, physiological and disease states and cell life in general. We devised three data fusion-based models for the proteome-level prediction of PPIs, and we show that our method outperforms state of the art approaches on common benchmarks. Moreover, we investigate its predictions on newly published PPIs, showing that this new data has a clear shift in its underlying distributions and we thus train and test our models on this extended dataset. Supplementary data are available at Bioinformatics online.

A Shape-Constrained Neural Data Fusion Network for Health Index Construction and Residual Life Prediction.

With the rapid development of sensor technologies, multisensor signals are now readily available for health condition monitoring and remaining useful life (RUL) prediction. To fully utilize these signals for a better health condition assessment and RUL prediction, health indices are often constructed through various data fusion techniques. Nevertheless, most of the existing methods fuse signals linearly, which may not be sufficient to characterize the health status for RUL prediction. To address this issue and improve the predictability, this article proposes a novel nonlinear data fusion approach, namely, a shape-constrained neural data fusion network for health index construction. Especially, a neural network-based structure is employed, and a novel loss function is formulated by simultaneously considering the monotonicity and curvature of the constructed health index and its variability at the failure time. A tailored adaptive moment estimation algorithm (Adam) is proposed for model parameter estimation. The effectiveness of the proposed method is demonstrated and compared through a case study using the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) data set.

A generalized remaining useful life prediction method for complex systems based on composite health indicator

As one of the key techniques in Prognostics and Health Management (PHM), accurate Remaining Useful Life (RUL) prediction can effectively reduce the number of downtime maintenance and significantly improve economic benefits. In this paper, a generalized RUL prediction method is proposed for complex systems with multiple Condition Monitoring (CM) signals. A stochastic degradation model is proposed to characterize the system degradation behavior, based on which the respective reliability characteristics such as the RUL and its Confidence Interval (CI) are explicitly derived. Considering the degradation model, two desirable properties of the Health Indicator (HI) are put forward and their respective quantitative evaluation methods are developed. With the desirable properties, a nonlinear data fusion method based on Genetic Programming (GP) is proposed to construct a superior composite HI. In this way, the multiple CM signals are fused to provide a better prediction capability. Finally, the proposed integrated methodology is validated on the C-MAPSS data set of aircraft turbine engines.

Real time localization solution for land vehicle application using low-cost integrated sensors with GPS

The technique proposed in this research demonstrates a real time nonlinear data fusion solution based on extremely low-cost and grade inertial sensors for land vehicle navigation. Here, the utilized nonlinear multi-sensor data fusion (MSDF) is based on the combination between extremely low-cost micro electrical mechanical systems (MEMS) inertial, heading, pressure, and speed sensors in addition to satellite-based navigation system. The integrated navigation system fuses position and velocity states from the Global Positioning System (GPS), the velocity measurements from an odometer, heading angle observation from a magnetometer and navigation states from an inertial navigation system (INS). The implemented system performance is assessed through the post-processing of collected raw measurements and real time experimental work. The system that runs the real-time experiments is established on three connected platforms, two of them are based on a 32-bit ARMTM core and the third one is based 16-bit AVR ATMEL microcontroller. This microcontroller is connected to an on-board diagnostics (OBD) shield to collect the vehicle speed measurements. The raw data obtained from the integrated sensors is saved and post processed in MATLAB®. In normal conditions, the estimated position errors are reduced through the usage of INS/GPS integration with heading observation angle from a magnetometer. In GPS-denied environments, the integrated system uses the observations from INS, magnetometer in addition to the velocity from odometer to ensure a continuous and accurate navigation solution. A complementary filter (CF) is implemented to estimate and improve the pitch and roll angles calculations. In addition to that, an unscented Kalman filter (UKF) is used cascaded with the designed CF to complete the designed sensors fusion algorithm. Experimental results show that the designed MSDF can achieve a good level of accuracy and a continuous localization solution of a land vehicle in different GPS availability cases and can be implemented on the available in the market processors to be run in real time.

Linking Gaussian process regression with data-driven manifold embeddings for nonlinear data fusion.

In statistical modelling with Gaussian process regression, it has been shown that combining (few) high-fidelity data with (many) low-fidelity data can enhance prediction accuracy, compared to prediction based on the few high-fidelity data only. Such information fusion techniques for multi-fidelity data commonly approach the high-fidelity model f h(t) as a function of two variables (t, s), and then use f l(t) as the s data. More generally, the high-fidelity model can be written as a function of several variables (t, s 1, s 2….); the low-fidelity model f l and, say, some of its derivatives can then be substituted for these variables. In this paper, we will explore mathematical algorithms for multi-fidelity information fusion that use such an approach towards improving the representation of the high-fidelity function with only a few training data points. Given that f h may not be a simple function-and sometimes not even a function-of f l, we demonstrate that using additional functions of t, such as derivatives or shifts of f l, can drastically improve the approximation of f h through Gaussian processes. We also point out a connection with 'embedology' techniques from topology and dynamical systems. Our illustrative examples range from instructive caricatures to computational biology models, such as Hodgkin-Huxley neural oscillations.

High Resolution Defect Imaging in Guided Waves Inspections by Dispersion Compensation and Nonlinear Data Fusion

High Resolution Defect Imaging in Guided Waves Inspections by Dispersion Compensation and Nonlinear Data Fusion

Nondestructive detection of total volatile basic nitrogen (TVB-N) content in pork meat by integrating hyperspectral imaging and colorimetric sensor combined with a nonlinear data fusion

Total volatile basic nitrogen (TVB-N) content is an important indicator in evaluating pork meat's freshness. This paper attempted a new strategy for measurement of TVB-N content in the pork meat by integrating two nondestructive sensing tools of hyperspectral imaging (HSI) and colorimetric sensors. First, HSI system and colorimetric sensors array system were used for data acquisition and further data processing, respectively. Herein, we proposed a novel efficient back propagation adaptive boosting (BP-AdaBoost) algorithm for data fusion and modeling, and we compared it with the classic modeling algorithm based on principal component analysis and back propagation artificial neural network (PCA-BPANN). Experiments results showed that the model based on data fusion was superior to the model based on the single sensing tool, and BP-AdaBoost has stronger capacity in solution to the complicated data fusion in contrast the PCA-BPANN. And the optimum results were achieved with the ratio of prediction to deviation (RPD) = 2.885, and the correction coefficient (R) = 0.932 in the prediction set. This work demonstrates that it has the potential to nondestructive detection of TVB-N content in pork meat by integrating HSI technique and colorimetric sensors technique combined with BP-AdaBoost nonlinear data fusion algorithm.

Cubature Kalman Fusion For Bearings-Only Tracking Networks

In this paper, we study the nonlinear data fusion problem for a kind of Bearings-only tracking network system with correlations among measurement noises. As a result, a decentralized cubature Kalman fusion algorithm is proposed by using an information filtering form (CIF) of cubature Kalman filter (CKF) and a noise de-correlation way. Based on the CIF and an augmented measurement, a centralized cubature Kalman fusion algorithm is firstly established and a problem on computational performance is pointed out. In order to improve computational performance of the centralized fusion method, the noise de-correlation technology is used to obtain a functionally equivalent Bearings-only tracking fusion system without the noise correlations, namely the variance of an augmented measurement noise is diagonalized. Accordingly, the proposed decentralized fusion method based on the CIF can be used to achieve a more effective tracking fusion estimate.

Interpretation and Visualization of Non-Linear Data Fusion in Kernel Space: Study on Metabolomic Characterization of Progression of Multiple Sclerosis

BackgroundIn the last decade data fusion has become widespread in the field of metabolomics. Linear data fusion is performed most commonly. However, many data display non-linear parameter dependences. The linear methods are bound to fail in such situations. We used proton Nuclear Magnetic Resonance and Gas Chromatography-Mass Spectrometry, two well established techniques, to generate metabolic profiles of Cerebrospinal fluid of Multiple Sclerosis (MScl) individuals. These datasets represent non-linearly separable groups. Thus, to extract relevant information and to combine them a special framework for data fusion is required.MethodologyThe main aim is to demonstrate a novel approach for data fusion for classification; the approach is applied to metabolomics datasets coming from patients suffering from MScl at a different stage of the disease. The approach involves data fusion in kernel space and consists of four main steps. The first one is to extract the significant information per data source using Support Vector Machine Recursive Feature Elimination. This method allows one to select a set of relevant variables. In the next step the optimized kernel matrices are merged by linear combination. In step 3 the merged datasets are analyzed with a classification technique, namely Kernel Partial Least Square Discriminant Analysis. In the final step, the variables in kernel space are visualized and their significance established.ConclusionsWe find that fusion in kernel space allows for efficient and reliable discrimination of classes (MScl and early stage). This data fusion approach achieves better class prediction accuracy than analysis of individual datasets and the commonly used mid-level fusion. The prediction accuracy on an independent test set (8 samples) reaches 100%. Additionally, the classification model obtained on fused kernels is simpler in terms of complexity, i.e. just one latent variable was sufficient. Finally, visualization of variables importance in kernel space was achieved.

THE ATTITUDE DETERMINATION ALGORITHM USING INTEGRATED GPS/INS DATA Yi-Min Huang

Traditionally, the extended Kalman filter (EKF) is a well known approach for general nonlinear data fusion problem. But the unit norm constraint of quaternion vector, which is the most efficient expression of attitude, results in the singularity of error covariance matrix. To overcome this difficulty, the modified extended Kalman filter (MEKF) has been developed by Shuster (1982). In order to improve the computational complexity of MEKF, we introduce the concept of spherical linear interpolation (SLERP). SLERP is a simple approach when two quaternions are averaged according to their accuracy dependent weightings. The SLERP based MEKF is proposed in this paper. However, its shortcoming is that the SLERP based MEKF cannot estimate gyro bias. Hence the proposed approach is not appropriate for the long duration navigation. The trade-off of MEKF and the SLERP based MEKF is discussed also.

INS algorithm using quaternion model for low cost IMU

This paper presents a generic inertial navigation system (INS) error propagation model that does not rely on small misalignment angles assumption. The modelling uses quaternions in the computer frame approach. Based on this model, an INS algorithm is developed for low cost inertial measurement unit (IMU) to solve the initial attitudes uncertainty using in-motion alignment. The distribution approximation filter (DAF) is used to implement the non-linear data fusion algorithm.