Lower Estimation Error Research Articles

Parametric evaluation of Kelvin-Voigt dispersion curve fitting

Ultrasound shear wave elastography (SWE) is a useful technique for non-invasive tissue assessment. For effective quantitative analysis the employment of physical models is necessary for rheological parameter estimation, such as the Kelvin-Voigt (KV) model. From SWE data, the frequency response can be calculated, and the shear wave velocity dispersion can be fit, yielding the shear elasticity and shear viscosity, μ1 and μ2, respectively. In this study, a parametric evaluation is performed using staggered-grid finite-difference simulations of materials with μ1 = 1–25 kPa (increment: 1 kPa) and μ2 = 0–10 Pa s (increment: 0.5 Pa s), with an acoustic radiation force push of f-number equal to 2. The novel Stockwell transform combined with a slant frequency-wavenumber analysis (GST-SFK) was compared with the two-dimensional Fourier transform for dispersion curve estimation over a variety of frequency ranges. Regardless of dispersion curve estimation technique, the accuracy of estimating μ1 is confounded for μ2 values above 5 Pa s. It was found that KV fitting benefits from wider frequency ranges (up to 1kHz), with lower μ1 and μ2 estimation errors. The GST-SFK was the best dispersion curve calculation technique, yet the KV fitting process was found to be unreliable for parameter estimation in highly viscous media.

Waveform design for higher-resolution localization with MIMO radar

To improve the accuracy and resolution of angle estimation with multiple signal classification (MUSIC) algorithm, we study the design of low peak-to-average-power-ratio waveforms for multiple-input-multiple-output (MIMO) radar. Assuming that the prior knowledge about the target directions and amplitudes is available (e.g., by transmitting initial probing waveforms), we formulate a waveform design problem based on minimizing the asymptotic mean square error of MUSIC. An alternating direction method of multipliers is developed to tackle the encountered nonconvex optimization problem. Compared with the competing waveforms, the waveforms designed by the proposed algorithm exhibit higher resolution and lower estimation error.

Productivity Index Prediction for Single-Lateral and Multilateral Oil Horizontal Wells Using Machine Learning Techniques.

Horizontal wells are geometrically shaped differently and projected to different flow regimes than vertical wells. Therefore, the existing laws that govern flow and productivity in vertical wells are not applicable to horizontal wells directly. The objective of this paper is to develop machine learning models that predict well productivity index using several reservoir and well inputs. Six models were developed using the actual well rate data from several wells divided into single-lateral wells, multilateral wells, and a combination of single-lateral and multilateral wells. The models are generated using artificial neural networks and fuzzy logic. The inputs used to create the models are the typical inputs used in the correlations and are well-known for any well under production. The results of the established ML models were excellent as suggested by an error analysis performed, reflecting the models to be robust. The error analysis showed high correlation coefficient values (between 0.94 and 0.95) supported by a low estimation error for four models out of six. The added value of this study is the developed general and accurate PI estimation model that overcomes many limitations of several widely used correlations in the industry and can be utilized for single-lateral or multilateral wells.

Multicompartment body composition analysis in older adults: a cross-sectional study

BackgroundDuring aging, changes occur in the proportions of muscle, fat, and bone. Body composition (BC) alterations have a great impact on health, quality of life, and functional capacity. Several equations to predict BC using anthropometric measurements have been developed from a bi-compartmental (2-C) approach that determines only fat mass (FM) and fat-free mass (FFM). However, these models have several limitations, when considering constant density, progressive bone demineralization, and changes in the hydration of the FFM, as typical changes during senescence. Thus, the main purpose of this study was to propose and validate a new multi-compartmental anthropometric model to predict fat, bone, and musculature components in older adults of both sexes.MethodsThis cross-sectional study included 100 older adults of both sexes. To determine the dependent variables (fat mass [FM], bone mineral content [BMC], and appendicular lean soft tissue [ALST]) whole total and regional dual-energy X-rayabsorptiometry (DXA) body scans were performed. Twenty-nine anthropometric measures and sex were appointed as independent variables. Models were developed through multivariate linear regression. Finally, the predicted residual error sum of squares (PRESS) statistic was used to measure the effectiveness of the predicted value for each dependent variable.ResultsAn equation was developed to simultaneously predict FM, BMC, and ALST from only four variables: weight, half-arm span (HAS), triceps skinfold (TriSK), and sex. This model showed high coefficients of determination and low estimation errors (FM: R2adj: 0.83 and SEE: 3.16; BMC: R2adj: 0.61 and SEE: 0.30; ALST: R2adj: 0.85 and SEE: 1.65).ConclusionThe equations provide a reliable, practical, and low-cost instrument to monitor changes in body components during the aging process. The internal cross-validation method PRESS presented sufficient reliability in the model as an inexpensive alternative for clinical field use.

Iterative algorithm for accurate superposition of contours with non-uniform sampling step

In this article, we describe an iterative algorithm for accurate superposition of contours with non-uniform sampling step. The processing contours are characterized by the same shape, but the sampling step is non-uniform, with no matching between points of the superposed contours. This makes impossible the use of methods for estimating superposition parameters by matching points. The algorithm proposed herein allows estimating the offsets and rotation angle separately. The idea of the algorithm is to perform the iterative correction of parameters. An estimate of the offsets is used to estimate the rotation angle and, vice versa, an estimate of the rotation angle is used to estimate the offsets. The proposed algorithm is characterized by a higher speed of processing than a brute force algorithm and a lower estimation error than algorithms that analyze contour macroparameters.

On the Importance of Non-Gaussianity in Chlorophyll Fluorescence Imaging

We propose a mathematical study of the statistics of chlorophyll fluorescence indices. While most of the literature assumes Gaussian distributions for these indices, we demonstrate their fundamental non-Gaussian nature. Indeed, while the noise in the raw fluorescence images can be assumed as Gaussian additive, the deterministic ratio between them produces nonlinear non-Gaussian distributions. We investigate the states in which this non-Gaussianity can affect the statistical estimation when wrongly approached with linear estimators. We provide an expectation–maximization estimator adapted to the non-Gaussian distributions. We illustrate the interest of this estimator with simulations from images of chlorophyll fluorescence indices.. We demonstrate the benefits of our approach by comparison with the standard Gaussian assumption. Our expectation–maximization estimator shows low estimation errors reaching seven percent for a more pronounced deviation from Gaussianity compared to Gaussianity assumptions estimators rising to more than 70 percent estimation error. These results show the importance of considering rigorous mathematical estimation approaches in chlorophyll fluorescence indices. The application of this work could be extended to various vegetation indices also made up of a ratio of Gaussian distributions.

State of Health Estimation for Smart Batteries using Transfer Learning with Data Cleaning

Smart battery with optimized pulsed current produced by the bypass is one prospective technology to prolong the service life of the batteries in electric vehicles. The accurate and reliable state of health (SOH) estimation is one significant process before the decision of the control. This paper proposes a proper solution for battery SOH estimation that can be applied to both constant and pulsed current charging scenarios. Specifically, a data cleaning process is proposed for preprocessing the fluctuated measurement, while retaining the main aging information. From the pre-processed data under different charging profiles, four SOH features are extracted, and the correlation coefficients prove their effectiveness with both constant current and pulsed currents. Later, a transfer learning-based model is developed which shows improved accuracy of the SOH estimations under pulsed current scenarios. Finally, experiments have been conducted to verify the proposed method. In the case of model retraining using only the first 10% of unseen data, satisfactory results can be obtained (the error is less than 2.626%). By increasing the data for model retraining to 20%, a fitted coefficient of larger than 0.994 between the estimations and real values is obtained, resulting in a low estimation error of less than 0.8%.

Fourier-transform infrared spectroscopy for characterization of liquid protein solutions: A comparison of two sampling techniques

Whereas industrial laboratory systems for Fourier-Transform Infrared (FTIR) spectroscopic characterization have been available for decades, dedicated FTIR solutions for in-line process control or for industrial at-line analysis of protein quality are still scarce. Thus, the present study aimed to compare two industrially viable sampling techniques, namely attenuated total reflectance (ATR) and dry film FTIR spectroscopy, for qualitative and quantitative analysis of liquid protein solutions. For this purpose, two sample sets were acquired: set 1 consisted of 95 protein hydrolysate samples produced in the laboratory from salmon processing by-products, and set 2 consisted of 133 protein hydrolysate samples obtained from an industrial processing plant of poultry by-products. Average molecular weights (AMW) and Brix measurements of protein hydrolysates were used as reference values for obtaining regression models. AMW was predicted with higher accuracy and lower estimation errors using dry film FTIR compared to ATR measurements for both sample types. The difference in predictive performance was higher in poultry hydrolysates because the protein and tissue complexities are higher than in salmon hydrolysates, and information from the amide I region in the FTIR spectra is needed to provide good calibration results. ATR, on the other hand, was more reliable for the prediction of Brix values of protein hydrolysates. The study also showed that FTIR spectroscopy can be used to predict protein quality features of industrially produced protein hydrolysates with a sensitivity of high industrial relevance. Therefore, developing industrially viable portable instruments for at-line process analysis in the food, feed, and biotech industries is a natural next development stage.

Unscented Kalman filter with stable embedding for simple, accurate, and computationally efficient state estimation of systems on manifolds in Euclidean space

AbstractThis article proposes a simple, accurate, and computationally efficient method to apply the ordinary unscented Kalman filter (UKF) developed in Euclidean space to systems whose dynamics evolve on manifolds. We use the mathematical theory called stable embedding to make a variant of UKF that keeps state estimates in close proximity to the manifold while exhibiting excellent estimation performance. We confirm the performance of our devised filter by applying it to the satellite system model and comparing the performance with other UKFs devised specifically for systems on manifolds. Our devised filter has a low estimation error, keeps the state estimates in close proximity to the manifold as expected, and consumes a minor amount of computation time. Also our devised filter is simple and easy to use because our filter directly employs the off‐the‐shelf standard UKF devised in Euclidean space without any particular manifold‐structure‐preserving discretization method or coordinate transformation.

An enhanced trilateration algorithm for indoor RSSI based positioning system using zigbee protocol

Location information of the object based on wireless communication will play crucial role in several Wireless Sensor Network (WSN) applications. Some applications need to know the exact position of the object. The advantages of Zigbee as the part of Radio Frequency (RF) technology such as low cost, high scalability, high availability, and supporting topology for Indoor positioning system (IPS). In this paper, we propose IPS using Received Signal Strength Indicator (RSSI) based Zigbee protocol. The proposed approach is based on the enhancement of Trilateration algorithm. Basically, the main concept of the conventional trilateration is using three strongest RSSI from references node. However, the instability from measured RSSI influenced to its estimation result. Therefore quadratic weighted is adding to this proposed scheme as the enhanced trilateration algorithm. The estimated distance output from conventional trilateration algorithm will be used for calculating the weighted value and multiplying to each the reference node which produced lower estimation error. The simulation result show that using enhanced trilateration algorithm has capability to improve accuracy of estimated position up to 90.55 % with mean square error (MSE) 2.03 meters compared with only using conventional trilateration reached high estimated error up to 4.31 meters.

Malaria Detection Accelerated: Combing a High-Throughput NanoZoomer Platform with a ParasiteMacro Algorithm.

Eradication of malaria, a mosquito-borne parasitic disease that hijacks human red blood cells, is a global priority. Microscopy remains the gold standard hallmark for diagnosis and estimation of parasitemia for malaria, to date. However, this approach is time-consuming and requires much expertise especially in malaria-endemic countries or in areas with low-density malaria infection. Thus, there is a need for accurate malaria diagnosis/parasitemia estimation with standardized, fast, and more reliable methods. To this end, we performed a proof-of-concept study using the automated imaging (NanoZoomer) platform to detect the malarial parasite in infected blood. The approach can be used as a steppingstone for malaria diagnosis and parasitemia estimation. Additionally, we created an algorithm (ParasiteMacro) compatible with free online imaging software (ImageJ) that can be used with low magnification objectives (e.g., 5×, 10×, and 20×) both in the NanoZoomer and routine microscope. The novel approach to estimate malarial parasitemia based on modern technologies compared to manual light microscopy demonstrated 100% sensitivity, 87% specificity, a 100% negative predictive value (NPV) and a 93% positive predictive value (PPV). The manual and automated malaria counts showed a good Pearson correlation for low- (R2 = 0.9377, r = 0.9683 and p < 0.0001) as well as high- parasitemia (R2 = 0.8170, r = 0.9044 and p < 0.0001) with low estimation errors. Our robust strategy that identifies and quantifies malaria can play a pivotal role in disease control strategies.

An Algorithm for Offline Measurement of Motor Stator Resistance and Voltage Drop Across Inverter Switches for Washing Machine Drives

The knowledge of the exact value of motor phase resistance is extremely important for modern electrical drives, which use it for tuning of current controllers, position estimators, temperature monitoring, and so on. At the same time, the value of voltage drop across inverter switches is necessary for providing proper and efficient control, especially in the low-speed region, where the impact of system nonlinearities is significant. This article proposes a novel algorithm for the offline measurement of motor resistance and voltage drop across inverter switches, which significantly improved the performance of washing machines in open-loop starting and low-speed operation. The proposed technique uses two step injection of dc current into stator windings, which provides data for the estimation of resistance and voltage drop. This article describes traps and pitfalls on the way of implementation and tuning and provides recommendations on avoiding them. After analysis of detailed experimental results, the author gives recommendation on a selection of algorithm parameters in order to provide low estimation errors. Following successful implementation and a series of tests, this technique has been accepted for mass production and has become an inevitable part of the control system of washing machines.

Technical indicators and cross-sectional expected returns

This study shows that 14 widely documented technical indicators explain cross-sectional stock returns. These indicators have lower estimation errors than the three-factor Fama–French and historical mean models. The long-short portfolios based on the cross-sectional technical signals generate substantial excess returns. These remain consistent after controlling for well-known cross-sectional return determinants, including momentum, size, book-to-market ratio, investment, and profitability. Our findings suggest that technical indicators play an important role in determining variation in cross-sectional returns.

Novel Iterative Machine Learning for Accurate Photon-Counting Poisson Channel Estimation

In this work, we present a new machine learning-based framework for the accurate estimation of the shot-noise limited photon-counting Poisson channel by developing a state-of-art iterative unsupervised learning algorithm for intelligent optical communications. By accurate estimation, we mean that it achieves very low estimation error even when the effect of the received data is unpredictable. We consider a realistic situation where modulated symbols are assumed to be hidden or unobserved latent variables, thereby making conventional estimation algorithms based on maximum likelihood approach unsuitable or inefficient. In particular, we consider a probabilistic model and assume that the received data is not labeled. With this unpredictable data considered, a novel iterative machine learning framework is developed based on an expectation and maximization algorithm. The proposed algorithm avoids the need to choose an appropriate step size as required in gradient method based algorithms. It is shown that it significantly outperforms the least square and the Viterbi detection technique.

Integrating spaceborne LiDAR and Sentinel-2 images to estimate forest aboveground biomass in Northern China

BackgroundFast and accurate forest aboveground biomass (AGB) estimation and mapping is the basic work of forest management and ecosystem dynamic investigation, which is of great significance to evaluate forest quality, resource assessment, and carbon cycle and management. The Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), as one of the latest launched spaceborne light detection and ranging (LiDAR) sensors, can penetrate the forest canopy and has the potential to obtain accurate forest vertical structure parameters on a large scale. However, the along-track segments of canopy height provided by ICESat-2 cannot be used to obtain comprehensive AGB spatial distribution. To make up for the deficiency of spaceborne LiDAR, the Sentinel-2 images provided by google earth engine (GEE) were used as the medium to integrate with ICESat-2 for continuous AGB mapping in our study. Ensemble learning can summarize the advantages of estimation models and achieve better estimation results. A stacking algorithm consisting of four non-parametric base models which are the backpropagation (BP) neural network, k-nearest neighbor (kNN), support vector machine (SVM), and random forest (RF) was proposed for AGB modeling and estimating in Saihanba forest farm, northern China.ResultsThe results show that stacking achieved the best AGB estimation accuracy among the models, with an R2 of 0.71 and a root mean square error (RMSE) of 45.67 Mg/ha. The stacking resulted in the lowest estimation error with the decreases of RMSE by 22.6%, 27.7%, 23.4%, and 19.0% compared with those from the BP, kNN, SVM, and RF, respectively.ConclusionCompared with using Sentinel-2 alone, the estimation errors of all models have been significantly reduced after adding the LiDAR variables of ICESat-2 in AGB estimation. The research demonstrated that ICESat-2 has the potential to improve the accuracy of AGB estimation and provides a reference for dynamic forest resources management and monitoring.

A High Accuracy & Ultra-Low Power ECG-Derived Respiration Estimation Processor for Wearable Respiration Monitoring Sensor.

The respiratory rate is widely used for evaluating a person’s health condition. Compared to other invasive and expensive methods, the ECG-derived respiration estimation is a more comfortable and affordable method to obtain the respiration rate. However, the existing ECG-derived respiration estimation methods suffer from low accuracy or high computational complexity. In this work, a high accuracy and ultra-low power ECG-derived respiration estimation processor has been proposed. Several techniques have been proposed to improve the accuracy and reduce the computational complexity (and thus power consumption), including QRS detection using refractory period refreshing and adaptive threshold EDR estimation. Implemented and fabricated using a 55 nm processing technology, the proposed processor achieves a low EDR estimation error of 0.73 on CEBS database and 1.2 on MIT-BIH Polysomnographic Database while demonstrating a record-low power consumption (354 nW) for the respiration monitoring, outperforming the existing designs. The proposed processor can be integrated in a wearable sensor for ultra-low power and high accuracy respiration monitoring.

Heterogeneous Transfer Learning for Wi-Fi Indoor Positioning Based Hybrid Feature Selection.

This paper presents the application of heterogeneous transfer learning (HetTL) methods which consider hybrid feature selection to reduce the training calibration effort and the noise generated by fingerprint duplicates obtained from multiple Wi-Fi access points. The Cramer–Rao Lower Bound analysis (CRLB) was also applied to evaluate and estimate a lower limit for the variance of a parameter estimator used to analyze positioning performance. We developed two novel algorithms for feature selection in fingerprint-based indoor positioning problems (IPP) to enhance positioning performance in the target domain with the HetTL. The algorithms comprised two scenarios: (i) a principal component analysis-based approach (PCA-based) and (ii) a hybrid approach that takes both PCA and correlation effect analysis into account (hybrid scenario). Accordingly, a new feature vector was constructed by retaining only the most significant predictors, and the most efficient feature dimensions were also determined by using a hybrid-based approach. Experimental results showed that the hybrid-based proposed algorithm has the minimum mean absolute error. The CRLB analysis also showed that the number of Wi-Fi access points could affect the lower bound location estimation error; however, identifying the most significant predictors is an effective approach to improve positioning performance.

EMISSA (Exploring millimetre indicators of solar-stellar activity)

Context. An activity indicator, which can provide a robust quantitative mapping between the stellar activity and the physical properties of its atmosphere, is important in exploring the evolution of the observed active phenomena across main-sequence stars of different spectral types. Common activity indicators do provide qualitative correlations with physical properties such as Teff and the rotation period, among others. However, due to the large variability in their values, even for a single star, defining robust quantitative mappings between activity and physical properties is difficult. Millimetre (mm) wavelengths probe the different atmospheric layers within the stellar chromosphere, providing a tomographic view of the atmospheric dynamics. Aims. The project aims to define a robust activity indicator by characterising mm brightness temperature spectra (TB(ν)) of the cool main-sequence stars (Teff ∼ 5000–7000 K) compiled by Paper I in this series. The sample contains 13 stars, including the Sun. Methods. We derived the mm TB(ν) spectral indices (αmm) for cool stars, including the Sun, based on observations in the 30–1000 GHz range. The derived values for αmm are explored as a function of various physical parameters and empirical power-law functions were derived. We also compared αmm estimates with other activity indicators. Results. Despite the estimation errors, αmm values could distinguish the cool stars well, unlike common activity indicators. The low estimation errors on the derived trends of αmm vs. physical parameters suggest that αmm could be a robust activity indicator. Conclusions. We note that αmm, which is linked to chromospheric thermal stratification and activity in cool stars, can well distinguish and physically characterise the stars more robustly than common activity indicators. We emphasise the need for multi-frequency data across the mm band for stars, with a range of physical parameters and gathered at multiple epochs during their activity cycles. This will help to explore αmm in a statistically robust manner and to study the emergence of chromospheric heating on the main sequence.

Performance assessment of interpolation techniques for optimal areal rainfall–temperature estimation: the case of two contrasting river catchments, Akaki and Mille, in Ethiopia

Abstract In the topographic complex catchments, landscape features have a significant impact on the spatial prediction of rainfall and temperature. In this study, performance assessments were made of various interpolation techniques for the prediction of the spatial distribution of rainfall and temperature in the Mille and Akaki River catchments, Ethiopia, through an improved approach on selecting the auxiliary variables as a covariate. Two geostatistical interpolation techniques, ordinary kriging (OK) and kriging with external drift (KED), and one deterministic interpolation technique, inverse distance weighting (IDW), were tested through a leave-one-out cross-validation (LOOCV) procedure. The results indicated that using the multivariate geostatistical interpolation technique (KED) with the auxiliary variables as a covsariate outperformed the univariate geostatistical (OK) and deterministic (IDW) techniques for the spatial interpolation of sampled rainfall–temperature data in both contrasting catchments, Akaki and Mille, with the lowest estimation errors (e.g., for Mille annual mean rainfall: root mean square error=75.32, 77.34, 245.72, mean bias error=3.70, −33.18, −15.61, mean absolute error=67.99, 69.51, 192.64) using KED with the combination of elevation and easting as a covariate, IDW and OK, respectively. Thus, the study confirmed that the use of elevation and easting/northing coordinates as predictors in geostatistical interpolation techniques could significantly improve the spatial prediction of climatic variables.

Deep Gaussian processes and infinite neural networks for the analysis of EEG signals in Alzheimer’s diseases

Deep neural network models (DGPs) can be represented hierarchically by a sequential composition of layers. When the prior distribution over the weights and biases are independently identically distributed, there is an equivalence with Gaussian processes (GP) in the limit of an infinite net[1]work width. DGPs are non-parametric statistical models used to character[1]ize patterns of complex non-linear systems due to their flexibility, greater generalization capacity, and a natural way of making inferences about the parameters and states of the system. This article proposes a hierarchi[1]cal Bayesian structure to model the weights and biases of a deep neural network. We deduce a general formula to calculate the integrals of Gaussian processes with non-linear transfer densities and obtain a kernel to estimate the covariance functions. In the methodology, we conduct an empirical study analyzing an electroencephalogram (EEG) database for diagnosing Alzheimer’s disease. Additionally, the DGPs models are esti[1]mated and compared with the NN models for 5, 10, 50, 100, 500, and 1000 neurons in the hidden layer, considering two transfer functions: Recti[1]fied Linear Unit (ReLU) and hyperbolic Tangent (Tanh). The results show good performance in the classification of the signals. Finally, we use the mean square error as a goodness of fit measure to validate the proposed models, obtaining low estimation errors.