Least-squares Linear Regression Research Articles

Radiometric Cross Calibration and Validation Using 4 Angle BRDF Model between Landsat 8 and Sentinel 2A

This work describes a proposed radiometric cross calibration between the Landsat 8 Operational Land Imager (OLI) and Sentinel 2A Multispectral Instrument (MSI) sensors. The cross-calibration procedure involves (i) correction of the MSI data to account for spectral band differences with OLI and (ii) normalization of Bidirectional Reflectance Distribution Function (BRDF) effects in the data from both sensors using a new model accounting for the view zenith/azimuth angles in addition to the solar zenith/view angles. Following application of the spectral and BRDF normalization, standard least-squares linear regression is used to determine the cross-calibration gain and offset in each band. Uncertainties related to each step in the proposed process are determined, as is the overall uncertainty associated with the complete processing sequence. Validation of the proposed cross-calibration gains and offsets is performed on image data acquired over the Algodones Dunes site. The results of this work indicate that the blue band has the most significant offset, requiring use of the estimated cross-calibration offset in addition to the estimated gain. The highest difference was observed in the blue and red bands, which are 2.6% and 1.4%, respectively, while other bands shows no significant difference. Overall, the net uncertainty in the proposed process was estimated to be on the order of 6.76%, with the largest uncertainty component due to each sensor’s calibration uncertainty on the order of 5% and 3% for the MSI and OLI, respectively. Other significant contributions to the uncertainty include seasonal changes in solar zenith and azimuth angles, target site nonuniformity, variability in atmospheric water vapor, and/or aerosol concentration.

Methodology to derive well-calibrated thermometers: a new glass-composition geothermometer for olivine-bearing glassy samples at one atmosphere

This work describes the methodology used to develop a new glass-composition geothermometer by multiple linear least-squares regression analysis of experimental data available in literature. The geothermometer is applicable to natural olivine-bearing glassy samples quenched at one atmosphere. The calibration data set includes a total of 543 anhydrous experiments performed at one atmosphere which produced olivine coexisting with glass (± other phases). The accuracy of the geothermometer has been verified using an independent set of randomly selected one atmosphere anhydrous experimental samples external to the calibration data set. The new proposed equation reads: T(°C) = 1054.1±6.5 + (1458.0±23.5)(XMgO)liq – (267.1±45.8)(XCaO)liq + (116.3±30.9)(XNa2O + XK2O)liq, where the terms such as (XMgO)liq represent the cation fraction of MgO in the liquid. The new geothermometer is able to reproduce the calibration temperatures within a standard error of estimate (SEE) of ±23° C and to predict the temperatures for the test data set with a SEE of ±25° C. The proposed equation, when compared to previous models, gives the lowest systematic error. Moreover, we mathematically demonstrate that the comparison of the slope and the intercept parameters of the regression line against the 1:1 line leads to correct evaluation of the model only when a regression of observed or measured (in the y-axis as dependent variable) vs. predicted or calculated (in the x-axis as independent variable) is used.

Impact of the 2005 and 2010 Spanish smoking laws on hospital admissions for tobacco-related diseases in Valencia, Spain

ObjectivesThis study aimed to assess the impact of the latest smoke-free legislation on hospital admission rates due to smoking-related diseases in Spain. Study designA retrospective cohort study was conducted to evaluate changes in hospital admission rates for cardiovascular, respiratory diseases, and smoking-related cancer in Valencia, Spain, during the period 1995–2013. Law 28/2005 and then law 42/2010 prohibited smoking in bars and restaurants as well as playgrounds and access points to schools and hospitals. MethodsGeneral population data by age and sex were obtained from the National Institute of Statistics census. Data on hospital admissions were obtained from the Minimum Basic Data Set. Diagnoses were codified according to the International Classification of Diseases-9th revision. Data from all hospitals of the Valencian Community from 1995 to 2013 were analysed. Trend analyses in the periods before and after the approval of the 2005 law were conducted using least-squares linear regression models. ResultsAdjusted hospital admission rates per 100,000 inhabitants for cardiovascular diseases significantly decreased after the 2005 Law (from 550.0/100,000 in 2005 to 500.5/100,000 in 2007), with a further decrease (to 434.6/100,000) in 2013, after the 2010 Law. Reductions in hospital admissions were seen in men and women, although declining trends were more marked in men. Hospital admission rates for respiratory diseases showed a reduction of a lower magnitude, whereas for smoking-related cancer admissions there was a slight decline only among men. ConclusionsThe Spanish comprehensive smoking ban resulted in a remarkable reduction of the adjusted rate of hospital admissions mainly associated to cardiovascular diseases. The decrease in the number of persons requiring in-patient care is relevant and may be viewed as an improvement of the public's health.

A validated LC–MS/MS method for the quantification of amlodipine, bisoprolol, enalapril and enalaprilat—Application to pharmacokinetic study in healthy volunteers

In this work, a highly sensitive and rapid liquid chromatography-tandem mass spectrometry (LC–MS/MS) method for the simultaneous determination of amlodipine (AML), bisoprolol (BIS) and enalapril maleate (ENA) in the presence of its metabolite enalaprilat (ENT) in real human plasma was developed and validated using felodipine, propranolol, ramipril, and captopril, as internal standards (IS), respectively. The prepared samples were chromatographed using Eclipse C18 column (4.6 × 100 mm, 5 μm) and the mobile phase was pumped in an isocratic mode consisting of acetonitrile: 0.01% formic acid (70:30, v/v) at a flow rate 0.7 mL min−1. The detection was achieved on an API 4500 triple quadrupole mass spectrometer using multiple reaction monitoring (MRM) mode in the positive electrospray ionization interface. Linearity was obtained over a concentration range of 0.1–10, 0.5–50, 5–500 and 1–100 ng/mL for AML, BIS, ENA and ENT respectively, by applying weighted least-squares linear regression method (1/x2). The accuracy and sensitivity of the proposed method allowed its application to determine the pharmacokinetics of these analytes in human male volunteers following a single oral administration.

Involvement of NO in Antinociception of NSAIDS in Murine Formalin Hind Paw Assay.

There are different animal models to evaluate pain among them the formalin hind paw assay which is widely used since some of its events appear to be similar to the clinical pain of humans. The assay in which a dilute solution of formalin is injected into the dorsal hindpaw of a murine produces two 'phases' of pain behavior separated by a inactive period. The early phase (Phase I) is probably due to direct activation of nociceptors and the second phase (Phase II) is due to ongoing inflammatory input and central sensitization. Mice were used to determine the potency antinociceptive of piroxicam (1,3,10,and 30 mg/kg), parecoxib (0.3, 1,3,10 and 30 mg/kg), dexketoprofen (3,10,30 and 100 mg/kg) and ketoprofen (3,10,30 and 100 mg/kg). Dose-response for each NSAIDs were created before and after 5 mg/kg of L-NAME i.p. or 5 mg/kg i.p. of 7-nitroindazole. A least-squares linear regression analysis of the log dose-response curves allowed the calculation of the dose that produced 50% of antinociception (ED50) for each drug. The ED50 demonstrated the following rank order of potency, in the phase I: piroxicam > dexketoprofen > ketoprofen > parecoxib and in the phase II: piroxicam > ketoprofen > parecoxib > dexketoprofen. Pretreatment of the mice with L-NAME or 7-nitroindazol induced a significant increase of the analgesic power of the NSAIDs, with a significant reduction of the ED50. It is suggested that NO may be involved in both phases of the trial, which means that nitric oxide regulates the bioactivity of NSAIDs.

Description and Attribution Analysis of the 2017 Spring Anomalous High Temperature Causing Floods in Kazakhstan

It is speculated that floods in many areas of the world have become more severe with global warming. This study describes the 2017 spring floods in Kazakhstan, which, with about six people dead or missing, prompted the government to call for more than 7,000 people to leave their homes. Then, based on the Climatic Research Unit (CRU), the NCEP/NCAR Reanalysis 1, and the Coupled Model Intercomparison Project 5 (CMIP5) simulations, the seasonal trends of temperature were calculated using the linear least-squares regression and the Mann–Kendall trend test. The correlation between the surface air temperature and atmospheric circulation was explored, and the attributable risk of the 2017 spring floods was evaluated using the conventional fraction of the attributable risk (FAR) method. The results indicate that the north plains of Kazakhstan had a higher (March–April) mean temperature anomaly compared to the south plains, up to 3°C, relative to the 1901-2017 average temperature. This was the primary cause of flooding in Kazakhstan. March and April were the other months with a higher increasing trend in temperature from 1901 to 2017 compared with other months. In addition, a positive anomaly of the geopotential height and air temperature for the March–April 2017 period (based on the reference period 1961-1990) was the reason for a warmer abnormal temperature in the northwest region of Kazakhstan. Finally, the FAR value was approximately equal to 1, which supported the claim of a strong anthropogenic influence on the risk of the 2017 March–April floods in Kazakhstan. The results presented provide essential information for a comprehensive understanding of the 2017 spring floods in Kazakhstan and will help government officials identify flooding situations and mitigate damage in future.

Effect of Firing Cycle and Etching Protocols on Tensile Bond Strength of Composite Cement to Zirconium-incorporated Lithium-Silicate Glass Ceramic.

Purpose: To evaluate whether the composite cement tensile bond strength is affected by different firing cycles or etching conditions on zirconium-incorporated lithium-silicate glass ceramics. Materials and Methods: Specimens (N=53) were prepared from blocks of zirconium-incorporated lithium-silicate glass ceramic (ZLS, Dentsply Sirona) with specimen dimensions of 12.5 x 14 x 2 mm (L x W x H, respectively). The specimens were subjected to different firing cycles and etching conditions. They were subsequently cemented onto titanium rods (grade V Ti-alloy, 4.8 mm in diameter, and 25.4 mm in length) with self-adhesive composite cement (TheraCem, Bisco), and then tested for tensile bond strength (TBS). The least-square means linear regression model was used to analyze the effects on TBS using JMP Pro 14 (SAS) and the post-hoc Tukey test with α = 0.05. Results: The TBS was significantly affected by etching duration (p < 0.001) and firing cycles (p < 0.001), but was not significantly affected by etchant concentration (p = 0.31). The highest TBS peak (4.83 MPa) on the fit curve was observed at 45 s of etching time. The lowest TBS was observed at 20 s (3.02 ± 0.28 MPa) etching time. In terms of firing cycles, 1 firing cycle (4.00 ± 0.71 MPa) provided the highest TBS. In contrast, the lowest TBS was observed for as-machined ZLS specimens but without a firing step (3.29 ± 0.53 MPa). Conclusions: Changing the etching time and firing conditions of the specimens had a significant effect on the composite cement tensile bond strength to zirconium-incorporated lithium-silicate glass ceramic. In contrast, etching the specimens with different acid concentrations, 5% or 9.6% of HF, did not significantly change the composite cement tensile bond strength.

An efficient and reliable method for determining the angle of repose of biomass by using 3D scan

This paper proposed a novel methodology to determine the angle of repose (AoR) of various biomass materials. Because of irregular shape and varying particle size distribution, slopes of biomass materials are uneven frequently and the two traditional measurement methods of AoR are therefore less applicable and reliable. In general, only partial slopes of heap surface are selected to measure AoR, although a representable slope is barely defined in standard methods. In this case, the subjective selection of slopes may result in significant deviations. Hence, we presented an efficient and reliable measurement method of AoR. In addition to the traditional bottomless cylinder test, the new technology, 3D scan, was used to digitize the generated bulk heap. Then, two data processing methods were introduced to deal with different heap forms: direct calculation based on the linear least-squares regression with regard to the whole heap surface and the pre-selection of planes for excluding outliners caused by particle bridging. For the purpose of simplifying the process, a self-developed graphic user interface (GUI) was used to analyze four kinds of biomass fuels, namely, olive stone, almond shells, forest residues and willow chips, and verify the proposed method. Both methods significantly decreased the deviations of results without increasing operating time and cost.

Effects of micronutrients supplementation on plasma homocysteine level and global cognitive function among older adults in long-term care facilities: a randomized controlled study

Objective To investigate the effects of micronutrient supplementation for the elderly on plasma homocysteine level and cognitive function in institutional older adults. Methods A total of 98 older adults with the score≤11 by mini nutritional assessment short-form aged 65-100 years were enrolled and assigned to either intervention group or control group (n=49 each), with either a package of micronutrient pack or placebo daily, for three months. Fasting blood samples were collected both at baseline and the end of study to detect serum vitamin B12, folate and plasma homocysteine (Hcy) concentrations. Global cognitive function was assessed using the mini-mental state examination (MMSE). The paired t test was used to compare the continuous variables between the two groups after intervention. The relationship between changes in MMSE score and changes in plasma Hcy concentrations was examined by least-squares linear regression. Results Eighty-two patients completed the study, and 17 patients withdrew from the study due to diarrhea and hospital discharge with the drop rate of 17.3%. Compared to the control group, concentrations of serum vitamin B12 (128.8±34.8 vs 13.3±16.0 pmol/L, P=0.003) and folate (21.1±1.6 vs 0.6±0.5 nmol/L, P<0.01) significantly increased in the intervention group over 3-month supplementation, while plasma Hcy levels were remarkably reduced (-5.3±0.7 vs 1.7±0.3 μmol/L, P<0.01). The incidences of deficiency of folate, deficiency of serum vitamin B12and high Hcy all decreased in intervention group. Although individual item scores in MMSE were not changed markedly, change of total MMSE score in intervention group were higher than that in the control group (1.2±3.0 vs -0.2±2.5, P<0.05). Conclusions The supplementation of the micronutrient pack in long-term care facilities can reduce the incidence of hyperhomocysteinemia, and improve the total MMSE score. Key words: Micronutrient; Older adults; Homocysteine; Cognitive function

Comparison of matrix-effect corrections for ordinary and uncertainty weighted linear regressions and determination of major element mean concentrations and total uncertainties of sixty-two international geochemical reference materials from wavelength-dispersive X-ray fluorescence spectrometry

In X-ray fluorescence (XRF) spectrometry, matrix-effect corrections for two types of linear regressions (ordinary and uncertainty-based weighted) have been performed for the first time and compared for the determination of major elements in rocks and minerals. The analytical data from different laboratories for 62 international geochemical reference materials (GRMs) were first processed to obtain better estimates of the mean values, along with lower uncertainties (narrower confidence limits of the mean), which were used in the XRF calibrations. The blank intensities were subtracted from the response for each GRM and the respective uncertainties were estimated. The uncertainty-based weighted least-squares linear regression (UWLR) model, being statistically more appropriate than the ordinary least-squares linear regression (OLR) model, provided more reliable regression equations than the OLR. All calculations and matrix-effect corrections were achieved through a newly developed online computer program. Similarly, matrix-effect corrections involving 11 α and 26 α (alphas) are presented and compared. The best UWLR method involving 26 α was successfully applied to the 62 GRMs and 4 similarly complex rock matrices as the calibrators. Both evaluations confirmed the usefulness of the UWLR model. This UWLR model was also compared with the OLR 26 α model for 62 GRMs treated as unknowns and shown to perform better. Thus, the UWLR model, along with the proposed matrix-effect correction method of 26 α, can be recommended as the most appropriate procedure for the calibration of XRF instruments, instead of the commonly used OLR models. This is the first time when 26 α are advantageously in a better estimate of influence coefficients used for major element determinations in rock and mineral samples. Because the UWLR calibration is based on a large number of GRMs of many rock and mineral types, it should be useful for all kinds of geological materials.

Comparison of linearization methods for modeling the Langmuir adsorption isotherm

The Langmuir isotherm model is widely used for modeling adsorption equilibrium data. The linearization method is mostly applied for estimations of the Langmuir parameters. However, it is not unified and requires further study. In this paper, the applicability and accuracy of four linearization methods of the Langmuir isotherm (equilibrium adsorbate concentration (Ce)/equilibrium adsorption capacity (qe)-Ce, 1/qe-1/Ce, qe-qe/Ce and qe/Ce-qe) in parameter estimations by linear least squares regression (LSR) were compared. The influence of linearization on the assumption about errors in the LSR variables was analyzed. The results showed that Langmuir-1 (Ce/qe-Ce) and Langmuir-4 (qe/Ce-qe) followed the assumption for all evaluated conditions whereas Langmuir-3 (qe-qe/Ce) violated the assumption for all conditions. Langmuir-2 (1/qe-1/Ce) obeyed the assumption of LSR only when Ce approached the initial adsorbate concentration (C0). Langmuir-1 (Ce/qe-Ce) provided the most accurate estimations of the Langmuir parameters, particularly at low levels of random errors in Ce and when the errors (%) in estimated maximum adsorption capacity (qm) and constant (Ka) ranged from −5% to 5% and from −15% to 15% at low levels of errors (−0.5–0.5 mg/L), respectively. The Langmuir-1 isotherm model adequately represented the experimental equilibrium data of sulfamethoxazole (SMX) onto activated carbon with an R2 value of 0.96.

Rapid in situ analysis of l-histidine and α-lactose in dietary supplements by fingerprint peaks using terahertz frequency-domain spectroscopy

A simple, green and nondestructive method based on terahertz fingerprint peaks has been developed for rapid in situ analysis of l-histidine and α-lactose in dietary supplements. Fingerprint absorption peaks of l-histidine and α-lactose located at 0.77 and 0.53 THz could be directly used for identification and quantitation of these analytes in commercial dietary supplements. Compared with the partial least squares regression model (PLSR), the linear least squares regression (LLSR) method based on peak areas presented better performance, with the linear correlation coefficients of 0.9899 and 0.9910 for l-histidine and α-lactose, respectively. Furthermore, analysis time per sample can be shortened to less than 1 min due to the narrower spectral acquisition region. The accuracies were 94.8–110% and 98.9–110%, comparable to those of ion chromatography for l-histidine and high-performance liquid chromatography for α-lactose. The results presented great potential of the developed method for rapid in situ analysis of nutrients in dietary supplements.

Rational Polynomial Chaos Expansions for the Stochastic Macromodeling of Network Responses

This paper introduces rational polynomial chaos expansions for the stochastic modeling of the frequency-domain responses of linear electrical networks. The proposed method models stochastic network responses as a ratio of polynomial chaos expansions, rather than the standard single polynomial expansion. This approach is motivated by the fact that network responses are best represented by rational functions of both frequency and parameters. In particular, it is proven that the rational stochastic model is exact for lumped networks. The model coefficients are computed via an iterative re-weighted linear least-square regression. Several application examples, concerning both lumped and a distributed systems, illustrate and validate the advocated methodology.

Water Demand Prediction for Housing Apartments Using Time Series Analysis

Water management has always been a topic of serious discussion since infrastructure, rural, and industrial development flourished. Due to the depleting water resources, this is now even a bigger challenge. So, here is developed an IoT-based water management system where ultrasonic sensors are employed for predicting the depth of water in the tank and accordingly pumping the water to the sub tank of the apartment. In addition, the time series analysis Auto Regressive Integrative Moving Average (ARIMA) and Least Square Linear Regression (LSLR) algorithms were employed and compared for predicting the water demand for next six months based on the historical water consumption record of the main reservoir/tank. The information on the amount of water consumed from the main reservoir is pushed to the cloud and to the mobile application developed for utilities. The purpose is to access the water consumption pattern and predict water demand for the next six months from the cloud.

Hyperspectral inversion of Suaeda salsa biomass under different types of human activity in Liaohe Estuary wetland in north-eastern China

Human activities alter the growth of coastal wetland vegetation. In the present study, we used a spectrometer and hyperspectral data to determine and compare the biomass of Suaeda salsa in a coastal wetland under protective and destructive activities. Using typical discriminants, the hyperspectral data of Suaeda salsa were distinguished under the influence of two kinds of human activity, and the accuracy of the inversion model of biomass was established following improved differentiation of the data under the influence of human activities. The original spectral reflectance and vegetation index were selected, and the biomass-inversion model was established by linear regression and partial least-squares regression. The model established by partial least-squares regression had a good precision (R2&amp;gt;0.85, RMSE%&amp;lt;5.6%). Hyperspectral technology can accurately show plant biomass and the indirect effects of interference by human activities of different intensity on coastal wetlands. The accuracy of the models can be improved by distinguishing the vegetation patterns under the influence of different types of human activity, and then constructing the biomass models. This study provides technical support for the use of quantitative remote sensing-based methods to monitor the fragile ecology of coastal wetlands under the influence of human activities.

Synergy of Satellite, In Situ and Modelled Data for Addressing the Scarcity of Water Quality Information for Eutrophication Assessment and Monitoring of Swedish Coastal Waters

Monthly CHL-a and Secchi Depth (SD) data derived from the full mission data of the Medium Resolution Imaging Spectrometer (MERIS; 2002–2012) were analysed along a horizontal transect from the inner Bråviken bay and out into the open sea. The CHL-a values were calibrated using an algorithm derived from Swedish lakes. Then, calibrated Chl-a and Secchi Depth (SD) estimates were extracted from MERIS data along the transect and compared to conventional monitoring data as well as to data from the Swedish Coastal zone Model (SCM), providing physico-biogeochemical parameters such as temperature, nutrients, Chlorophyll-a (CHL-a) and Secchi depth (SD). A high negative correlation was observed between satellite-derived CHL-a and SD (ρ = −0.91), similar to the in situ relationship established for several coastal gradients in the Baltic proper. We also demonstrate that the validated MERIS-based estimates and data from the SCM showed strong correlations for the variables CHL-a, SD and total nitrogen (TOTN), which improved significantly when analysed on a monthly basis across basins. The relationship between satellite-derived CHL-a and modelled TOTN was also evaluated on a monthly basis using least-square linear regression models. The predictive power of the models was strong for the period May-November (R2: 0.58–0.87), and the regression algorithm for summer was almost identical to the algorithm generated from in situ data in Himmerfjärden bay. The strong correlation between SD and modelled TOTN confirms that SD is a robust and reliable indicator to evaluate changes in eutrophication in the Baltic proper which can be assessed using remote sensing data. Amongst all three assessed methods, only MERIS CHL-a was able to correctly depict the pattern of phytoplankton phenology that is typical for the Baltic proper. The approach of combining satellite data and physio-biogeochemical models could serve as a powerful tool and value-adding complement to the scarcely available in situ data from national monitoring programs. In particular, satellite data will help to reduce uncertainties in long-term monitoring data due to its improved measurement frequency.

Corneal Biomechanics and Visual Field Progression in Eyes with Seemingly Well-Controlled Intraocular Pressure

To investigate the incidence and risk factors for glaucomatous visual field progression in eyes with well-controlled intraocular pressure (IOP). Prospective cohort. A total of 460 eyes of 334 patients with glaucoma under treatment. Study subjects had a mean follow-up of 4.3±0.8 years. Patients were classified as well controlled if all IOP measurements were less than 18 mmHg. Rates of visual field progression were calculated using ordinary least-squares linear regression of standard automated perimetry (SAP) mean deviation (MD) values over time. Progression was defined as a significantly negative MD slope (alpha= 0.05). Rates of SAP MD change; mean and peak IOP, and IOP fluctuation; and corneal biomechanics: corneal hysteresis (CH), central corneal thickness (CCT), and corneal index. Of the 179 eyes with well-controlled IOP, 42 (23.5%) demonstrated visual field progression. There was no significant difference between progressing and stable patients in baseline MD (-6.4±7.1 decibels [dB] vs.-6.0±6.2 dB; P= 0.346), mean IOP (11.7±2.0 mmHg vs. 12.1±2.3 mmHg; P= 0.405), IOP fluctuation (1.6±0.6 mmHg vs. 1.6±0.5 mmHg; P= 0.402), or peak IOP (14.3±1.9 mmHg vs. 14.6±2.1 mmHg; P= 0.926). Progressing eyes had significantly lower CH (8.6±1.3 mmHg vs. 9.4±1.6 mmHg; P= 0.014) and thinner CCT (515.1±33.1 μm vs. 531.1±42.4 μm; P= 0.018, respectively) compared with stable eyes. In the multivariate analysis, a 1 standard deviation lower corneal index, a summation of normalized versions of CH and CCT, resulted in a 68% higher risk of progression (odds ratio, 1.68; 95% confidence interval, 1.08-2.62; P= 0.021). Approximately one-quarter of eyes with well-controlled IOP may show visual field progression over time. Thin cornea and low CH are main risk factors.

Application of green analytical chemistry to a green chemistry process: Magnetic resonance and Raman spectroscopic process monitoring of continuous ethanolic fermentation.

Compact 1 H NMR and Raman spectrometers were used for real-time process monitoring of alcoholic fermentation in a continuous flow reactor. Yeast cells catalyzing the sucrose conversion were immobilized in alginate beads floating in the reactor. The spectrometers proved to be robust and could be easily attached to the reaction apparatus. As environmentally friendly analysis methods, 1 H NMR and Raman spectroscopy were selected to match the resource- and energy-saving process. Analyses took only a few seconds to minutes compared to chromatographic procedures and were, therefore, suitable for real-time control realized as a feedback loop. Both compact spectrometers were successfully implemented online. Raman spectroscopy allowed for faster spectral acquisition and higher quantitative precision, NMR yielded more resolved signals thus higher specificity. By using the software Matlab for automated data loading and processing, relevant parameters such as the ethanol, glycerol, and sugar content could be easily obtained. The subsequent multivariate data analysis using partial linear least-squares regression type 2 enabled the quantitative monitoring of all reactants within a single model in real time.

Attacking Data Transforming Learners at Training Time

While machine learning systems are known to be vulnerable to data-manipulation attacks at both training and deployment time, little is known about how to adapt attacks when the defender transforms data prior to model estimation. We consider the setting where the defender Bob first transforms the data then learns a model from the result; Alice, the attacker, perturbs Bob’s input data prior to him transforming it. We develop a general-purpose “plug and play” framework for gradient-based attacks based on matrix differentials, focusing on ordinary least-squares linear regression. This allows learning algorithms and data transformations to be paired and composed arbitrarily: attacks can be adapted through the use of the chain rule—analogous to backpropagation on neural network parameters—to compositional learning maps. Bestresponse attacks can be computed through matrix multiplications from a library of attack matrices for transformations and learners. Our treatment of linear regression extends state-ofthe-art attacks at training time, by permitting the attacker to affect both features and targets optimally and simultaneously. We explore several transformations broadly used across machine learning with a driving motivation for our work being autogressive modeling. There, Bob transforms a univariate time series into a matrix of observations and vector of target values which can then be fed into standard learners. Under this learning reduction, a perturbation from Alice to a single value of the time series affects features of several data points along with target values.

3D-Deep Learning Based Automatic Diagnosis of Alzheimer's Disease with Joint MMSE Prediction Using Resting-State fMRI.

We performed this research to 1) evaluate a novel deep learning method for the diagnosis of Alzheimer's disease (AD) and 2) jointly predict the Mini Mental State Examination (MMSE) scores of South Korean patients with AD. Using resting-state functional Magnetic Resonance Imaging (rs-fMRI) scans of 331 participants, we obtained functional 3-dimensional (3-D) independent component spatial maps for use as features in classification and regression tasks. A 3-D convolutional neural network (CNN) architecture was developed for the classification task. MMSE scores were predicted using: linear least square regression (LLSR), support vector regression, bagging-based ensemble regression, and tree regression with group independent component analysis (gICA) features. To improve MMSE regression performance, we applied feature optimization methods including least absolute shrinkage and selection operator and support vector machine-based recursive feature elimination (SVM-RFE). The mean balanced test accuracy was 85.27% for the classification of AD versus healthy controls. The medial visual, default mode, dorsal attention, executive, and auditory related networks were mainly associated with AD. The maximum clinical MMSE score prediction accuracy with the LLSR method applied on gICA combined with SVM-RFE features had the lowest root mean square error (3.27 ± 0.58) and the highest R2 value (0.63 ± 0.02). Classification of AD and healthy controls can be successfully achieved using only rs-fMRI and MMSE scores can be accurately predicted using functional independent component features. In the absence of trained clinicians, AD disease status and clinical MMSE scores can be jointly predicted using 3-D deep learning and regression learning approaches with rs-fMRI data.