Using Computer Vision and Intelligent Classification Techniques for the Classification and Selection of Brazilian Nuts

Aiming to improve the processes involved in the industrial beneficiation of the Brazilian nuts, this work used a new methodology based on concepts of computer vision and intelligent classification, with a focus on two of the various stages of the processing： classification according to the origin and selection. Using the proposed methodology for the selection of the nuts it was possible to distinguish between intact and broken nuts and between good and spoiled nuts with a very high percentage of correct identifications. Also to evaluate the efficiency of the proposed methodology, visual tests by human subjects were performed for the classification of the nuts, the results demonstrated that the intelligent techniques performed the same or better than the visual classification.

Heart Disease Risk Prediction Expending of Classification Algorithms

Heart disease prognosis(HDP)is a difficult undertaking that requires knowledge and expertise to predict early on.Heart failure is on the rise as a result of today’s lifestyle.The healthcare business generates a vast volume of patient records,which are challenging to manage manually.When it comes to data mining and machine learning,having a huge volume of data is crucial for getting meaningful information.Several methods for predictingHDhave been used by researchers over the last few decades,but the fundamental concern remains the uncertainty factor in the output data,aswell as the need to decrease the error rate and enhance the accuracy of HDP assessment measures.However,in order to discover the optimal HDP solution,this study compares multiple classification algorithms utilizing two separate heart disease datasets from the Kaggle repository and the University of California,Irvine(UCI)machine learning repository.In a comparative analysis,Mean Absolute Error(MAE),Relative Absolute Error(RAE),precision,recall,fmeasure,and accuracy are used to evaluate Linear Regression(LR),Decision Tree(J48),Naive Bayes(NB),Artificial Neural Network(ANN),Simple Cart(SC),Bagging,Decision Stump(DS),AdaBoost,Rep Tree(REPT),and Support Vector Machine(SVM).Overall,the SVM classifier surpasses other classifiers in terms of increasing accuracy and decreasing error rate,with RAE of 33.2631 andMAEof 0.165,the precision of 0.841,recall of 0.835,f-measure of 0.833,and accuracy of 83.49 percent for the dataset gathered from UCI.The SC improves accuracy and reduces the error rate for the Kaggle dataset,which is 3.30%for RAE,0.016 percent for MAE,0.984%for precision,0.984 percent for recall,0.984 percent for f-measure,and 98.44%for accuracy.

Comparative Study on Tool Fault Diagnosis Methods Using Vibration Signals and Cutting Force Signals by Machine Learning Technique

The state of cutting tool determines the quality of surface produced on the machined parts.A faulty tool produces poor sur face,inaccurate geometry and non-economic production.Thus,it is necessary to monitor tool condition for a.machining process to have superior quality and economic production.In the pre-sent study,fault classification of single point cutting tool for hard turning has been carried out by employing machine learning technique.Cutting force and vibration signals were acquired to monitor tool condition during machining.A set of four tooling conditions namely healthy,worn flank,broken insert and extended tool overhang have been considered for the study.The machine learning technique was applied to both vibration and cutting force signals.Discrete wavelet features of the signals have been extracted using discrete wavelet trans formation(DWT).This transformation represents a large dataset into approximation coeffcients which contain the most useful information of the dataset.Significant features,among features extracted,were selected using J48 decision tree technique.Clas-sification of tool conditions was carried out us ing Naive Bayes algorithm.A 10 fold cross validation was incorporated to test the validity of classifier.A comparison of performance of classifier was made between cutting force and vibration signal to choose the best signal acquisition method in classifying tool fault conditions using machine learning technique.

Investigating of Classification Algorithms for Heart Disease Risk Prediction

Prognosis of HD is a complex task that requires experience andexpertise to predict in the early stage. Nowadays, heart failure is rising dueto the inherent lifestyle. The healthcare industry generates dense records ofpatients, which cannot be managed manually. Such an amount of data is verysignificant in the field of data mining and machine learning when gatheringvaluable knowledge. During the last few decades, researchers have used differentapproaches for the prediction of HD, but still, the major problem is theuncertainty factor in the output data and also there is a need to reduce theerror rate and increase the accuracy of evaluation metrics for HDP. However,this study largess the comparative analysis of diverse classification algorithmsgoing on two different heart disease datasets taken from the Kaggle repositoryand University of California, Irvine (UCI) machine learning repository tofind the best solution for HDP. Going through comparative analysis, tenclassifiers;LR, J48, NB, ANN, SC, Bagging, DS, AdaBoost, REPT, and SVMare evaluated using MAE, RAE, precision, recall, f-measure, and accuracy.The overall finding indicates that for the dataset taken from UCI, the SVMclassifier performs well as compared to other classifiers in terms of increasingaccuracy and reducing error rate that is 33.2631 for RAE, and 0.165 forMAE, 0.841 for precision, 0.835 for recall, 0.833 for f-measure and 83.49%for accuracy. Whereas for dataset taken from Kaggle, the SC performs well interms of increasing accuracy and reducing error rate that is 3.30% for RAE,0.016 for MAE, 0.984 for precision, 0.984 for recall, 0.984 for f-measure, and98.44% for accuracy.

Using Landsat images to monitor changes in the snow-covered area of selected glaciers in northern Pakistan

Landsat satellite images were used to map and monitor the snow-covered areas of four glaciers with different aspects(Passu: 36.473°N, 74.766°E;Momhil: 36.394°N, 75.085°E; Trivor: 36.249°N,74.968°E; and Kunyang: 36.083°N, 75.288°E) in the upper Indus basin, northern Pakistan, from 1990-2014. The snow-covered areas of the selected glaciers were identified and classified using supervised and rule-based image analysis techniques in three different seasons. Accuracy assessment of the classified images indicated that the supervised classification technique performed slightly better than the rule-based technique. Snow-covered areas on the selected glaciers were generally reduced during the study period but at different rates. Glaciers reached maximum areal snow coverage in winter and premonsoon seasons and minimum areal snow coverage in monsoon seasons, with the lowest snow-covered area occurring in August and September. The snowcovered area on Passu glacier decreased by 24.50%,3.15% and 11.25% in the pre-monsoon, monsoon and post-monsoon seasons, respectively. Similarly, the other three glaciers showed notable decreases in snow-covered area during the pre-and post-monsoon seasons; however, no clear changes were observed during monsoon seasons. During pre-monsoon seasons, the eastward-facing glacier lost comparatively more snow-covered area than the westward-facing glacier. The average seasonal glacier surface temperature calculated from the Landsat thermal band showed negative correlations of-0.67,-0.89,-0.75 and-0.77 with the average seasonal snowcovered areas of the Passu, Momhil, Trivor and Kunyang glaciers, respectively, during pre-monsoon seasons. Similarly, the air temperature collected from a nearby meteorological station showed an increasing trend, indicating that the snow-covered area reduction in the region was largely due to climate warming.

Healthy individuals vs patients with bipolar or unipolar depression in gray matter volume

BACKGROUND Previous studies using voxel-based morphometry(VBM)revealed changes in gray matter volume(GMV)of patients with depression,but the differences between patients with bipolar disorder(BD)and unipolar depression(UD)are less known.AIM To analyze the whole-brain GMV data of patients with untreated UD and BD compared with healthy controls.METHODS Fourteen patients with BD and 20 with UD were recruited from the Mental Health Center of Shantou University between August 2014 and July 2015,and 20 nondepressive controls were recruited.After routine three-plane positioning,axial T2WI scanning was performed.The connecting line between the anterior and posterior commissures was used as the scanning baseline.The scanning range extended from the cranial apex to the foramen magnum.Categorical data are presented as frequencies and were analyzed using the Fisher exact test.RESULTS There were no significant intergroup differences in gender,age,or years of education.Disease course,age at the first episode,and Hamilton depression rating scale scores were similar between patients with UD and those with BD.Compared with the non-depressive controls,patients with BD showed smaller GMVs in the right inferior temporal gyrus,left middle temporal gyrus,right middle occipital gyrus,and right superior parietal gyrus and larger GMVs in the midbrain,left superior frontal gyrus,and right cerebellum.In contrast,UD patients showed smaller GMVs than the controls in the right fusiform gyrus,left inferior occipital gyrus,left paracentral lobule,right superior and inferior temporal gyri,and the right posterior lobe of the cerebellum,and larger GMVs than the controls in the left posterior central gyrus and left middle frontal gyrus.There was no difference in GMV between patients with BD and UD.CONCLUSION Using VBM,the present study revealed that patients with UD and BD have different patterns of changes in GMV when compared with healthy controls.

Does the ratio of the carpal tunnel inlet and outlet cross-sectional areas in the median nerve reflect carpal tunnel syndrome severity?

Although ultrasound measurements have been used in previous studies on carpal tunnel syndrome to visualize injury to the median nerve, whether such ultrasound data can indicate the severity of carpal tunnel syndrome remains controversial. The cross-sectional areas of the median nerve at the tunnel inlet and outlet can show swelling and compression of the nerve at the carpal. We hypothesized that the ratio of the cross-sectional areas of the median nerve at the carpal tunnel inlet to outlet accurately reflects the severity of carpal tunnel syndrome. To test this, high-resolution ultrasound with a linear array transducer at 5–17 MHz was used to assess 77 patients with carpal tunnel syndrome. The results showed that the cut-off point for the inlet-to-outlet ratio was 1.14. Significant differences in the inlet-to-outlet ratio were found among patients with mild, moderate, and severe carpal tunnel syndrome. The cut-off point in the ratio of cross-sectional areas of the median nerve was 1.29 between mild and more severe（moderate and severe） carpal tunnel syndrome patients with 64.7% sensitivity and 72.7% specificity. The cut-off point in the ratio of cross-sectional areas of the median nerve was 1.52 between the moderate and severe carpal tunnel syndrome patients with 80.0% sensitivity and 64.7% specificity. These results suggest that the inlet-to-outlet ratio reflected the severity of carpal tunnel syndrome.

Automated estimation of stellar fundamental parameters from low resolution spectra: the PLS method

PLS （Partial Least Squares regression） is introduced into an automatic estimation of fundamental stellar spectral parameters. It extracts the most correlative spectral component to the parameters （Teff, log g and [Fe/H]）, and sets up a linear regression function from spectra to the corresponding parameters. Considering the properties of stellar spectra and the PLS algorithm, we present a piecewise PLS regression method for estimation of stellar parameters, which is composed of one PLS model for Teff, and seven PLS models for log g and [Fe/H] estimation. Its performance is investigated by large experiments on flux calibrated spectra and continuum normalized spectra at different signal-to-noise ratios （SNRs） and resolutions. The results show that the piecewise PLS method is robust for spectra at the medium resolution of 0.23 nm. For low resolution 0.5 nm and 1 nm spectra, it achieves competitive results at higher SNR. Experiments using ELODIE spectra of 0.23 nm resolution illustrate that our piecewise PLS models trained with MILES spectra are efficient for O ～ G stars： for flux calibrated spectra, the systematic offsets are 3.8%, 0.14 dex, and -0.09 dex for Teff, log g and [Fe/H], with error scatters of 5.2%, 0.44 dex and 0.38 dex, respectively; for continuum normalized spectra, the systematic offsets are 3.8%, 0.12dex, and -0.13 dex for Teff, log g and [Fe/H], with error scatters of 5.2%, 0.49 dex and 0.41 dex, respectively. The PLS method is rapid, easy to use and does not rely as strongly on the tightness of a parameter grid of templates to reach high precision as Artificial Neural Networks or minimum distance methods do.