Wyniki wyszukiwania - BazTech

1

Automated malarial retinopathy detection using transfer learning and multi-camera retinal images

Rajendra Kurup Aswathy, Wigdahl Jeff, Benson Jeremy, Martínez-Ramón Manel, Solíz Peter, Joshi Vinayak

Biocybernetics and Biomedical Engineering

|

2023

|

Vol. 43, no. 1

109--123

EN

Cerebral malaria (CM) is a fatal syndrome found commonly in children less than 5 years old in Sub-saharan Africa and Asia. The retinal signs associated with CM are known as malarial retinopathy (MR), and they include highly specific retinal lesions such as whitening and hemorrhages. Detecting these lesions allows the detection of CM with high specificity. Up to 23% of CM, patients are over-diagnosed due to the presence of clinical symptoms also related to pneumonia, meningitis, or others. Therefore, patients go untreated for these pathologies, resulting in death or neurological disability. It is essential to have a low-cost and high-specificity diagnostic technique for CM detection, for which We developed a method based on transfer learning (TL). Models pre-trained with TL select the good quality retinal images, which are fed into another TL model to detect CM. This approach shows a 96% specificity with low-cost retinal cameras.

2

Evaluation of the NO2 concentration prediction possibility based on static and dynamic responses of TGS sensors at changing humidity levels

Kalinowski Paweł, Woźniak Łukasz, Jasiński Grzegorz, Jasiński Piotr

Metrology and Measurement Systems

|

2020

|

Vol. 27, nr 1

167--179

EN

The commercially available metal-oxide TGS sensors are widely used in many applications due to the fact that they are inexpensive and considered to be reliable. However, they are partially selective and their responses are influenced by various factors, e.g. temperature or humidity level. Therefore, it is important to design a proper analysis system of the sensor responses. In this paper, the results of examinations of eight commercial TGS sensors combined in an array and measured over a period of a few months for the purpose of prediction of nitrogen dioxide concentration are presented. The measurements were performed at different relative humidity levels. PLS regression was employed as a method of quantitative analysis of the obtained sensor responses. The results of NO2 concentration prediction based on static and dynamic responses of sensors are compared. It is demonstrated that it is possible to predict the nitrogen dioxide concentration despite the influence of humidity.

3

Application of Partial Regression Methods to Long Range Forecasts

Konca-Kędzierska K.

Przegląd Geofizyczny

|

2018

|

Z. 4

353--362

EN

The article presents the construction of a regression model for the long-range forecast of tercile categories of the monthly mean temperature. Two methods from the group of the partial least squares (PLS) and sparse partial least squares (SPLS) methods were used. The selected methods combine the properties of principal component analysis (PCA) with features of multiple regression methods, and apply the creation of latent layers. These methods also have no restrictions related to the independence of predictors and no constraints on the model dimension. The predictors are percentiles (10%, 50% and 90%) for selected fields of the NCEP/NCAR Reanalysis dataset. The model uses a time series of predictors for periods from 5 to 30 years. The obtained set of forecasts is subjected to the evaluation process based on indicators for the dependent period. This allows for the selection of a reliable ensemble of forecasts. The presented model was tested between January 2014 and December 2016.

4

An empirical examination of the thickness profile formation of twin-roll-cast magnesium strips

Kawalla C., Höck M., Ullmann M., Ringle C. M.

Archives of Civil and Mechanical Engineering

|

2018

|

Vol. 18, no. 1

227--234

EN

The recently developed technology of twin-roll-cast (TRC) magnesium strips permits an efficient production of magnesium sheets, primarily for the automotive industry. The focus of the paper is to develop a structural equation model explaining the variance of the thickness profile formation. Hence, the complex and partially unknown relationships between twin-roll casting process parameters and the thickness profile formation are analyzed using latent variables, e.g. the deformation resistance, length of contact arc, etc., which consist of several observed parameters. The fundamental process variables and their effect on the thickness profile formation during twin-roll casting are investigated and evaluated by partial least squares structural equation modeling (PLS-SEM) – a statistical method that fits networks of constructs to empirical data. The results of the predictive modeling technique allow an approximation of the existing interrelationships between thickness profiles, rolling force as well as processes in the roll gap which are typically difficult to measure directly using sensors. In this context, it was identified that the thickness profile variation is primarily caused by the forming force, which is mainly driven by the length of contact arc. Moreover, implications for the control of the thickness profile are derived.

5

Comparing CB-SEM and PLS-SEM : a case showing management accounting impact on performance

Siska L.

Polish Journal of Management Studies

|

2017

|

Vol. 15, No. 2

240--249

EN

The objective of this article is to contrast two broad approaches to structural equation modelling (SEM): covariance-based (CB-SEM) and variance-based partial least square (PLS-SEM). Each approach was applied to estimate parameters of the same case model. Even though the results reveal some numerical differences, these differences do not seem to be of a great practical importance and less restrictive assumptions speak in favour of PLS-SEM. This study is one of the first attempts to apply and compare both approaches to SEM on actual (and not simulated) data, in this case data on management accounting (MA) obtained from 101 Czech and Slovak companies. From managerial viewpoint, the final model demonstrates that adoption of strategic MA techniques themselves without increase in organizational capabilities is insufficient for achieving higher return-on-assets (ROA).

PL

Celem tego artykułu jest skontrastowanie dwóch szerokich podejść do modelowania równań strukturalnych (ang. structural equation modelling, SEM): opartego na kowariancji (ang. covariance-based structural equation modelling, CB-SEM) i na wariancie częściowego najmniejszego kwadratu (ang. partial least square, PLS-SEM). Każde podejście zastosowano do oszacowania parametrów tego samego modelowego przypadku. Mimo, że wyniki wykazują pewne różnice liczbowe, różnice te nie mają dużego znaczenia praktycznego, a mniej restrykcyjne założenia przemawiają za PLS-SEM. Badanie to jest jedną z pierwszych prób zastosowania i porównania obu podejść do SEM w odniesieniu do danych rzeczywistych (a nie symulowanych), w tym przypadku danych dotyczących rachunkowości zarządczej (ang. management accounting, MA) uzyskanych od 101 firm czeskich i słowackich. Z punktu widzenia zarządzania, ostateczny model pokazuje, że samodzielne wdrażanie strategii MA bez zwiększenia zdolności organizacyjnych nie wystarcza, aby osiągnąć wyższy zwrot z aktywów (ang. return-on-assets, ROA).

6

A novel stage-based KPLS-PLS monitoring and quality prediction approach for batch processes

Qi Y., Li Y., Sun Z.

Przegląd Elektrotechniczny

|

2013

|

R. 89, nr 1b

147--150

EN

A novel KPLS-PLS batch monitoring and quality prediction approach based on fuzzy clustering soft-partition is proposed to solve the stage-transition monitoring and prediction problem in multistage batch processes. The proposed method calculates firstly similarity indices between different time-slice data matrices of batch processes, then phase division algorithm is designed by fuzzy clustering based on the similarity index, following by a fuzzy membership grade transition identification step. By setting a series of KPLS and PLS models with time-varying covariance structures for transitions and steady phases, it reflects objectively the diversity of transitional characteristics, capture the nonlinear relationships among process variables of the transition and can monitor and predict batch processes more accurately and efficiently. The superiority of the proposed method is illustrated by applying it to industrial application of fed-batch penicillin fermentation process. The results clearly demonstrate the effectiveness and feasibility of the proposed method.

PL

Zaproponowano nową metodę KPLS ( kernel partial least squers) – PLS monitorowania i przewidywania wieloetapowych procesów wsadowych. Metoda oparta została o klastrowanie rozmyte, pozwala na wykrycie przejść między etapami i dokładniejsze przewidywanie przebiegu procesu przez uniknięcie wpływu nieliniowości. Wyższość proponowanej metody zilustrowano wykorzystując ją do badania przemysłowego procesu fermentacji wsadu pożywki penicyliny.

7

Modeling of heat source based on parameters of electron beam welding process

Lacki P., Adamus K., Wojsyk K., Zawadzki M., Nitkiewicz Z.

Archives of Metallurgy and Materials

|

2011

|

Vol. 56, iss. 2

455-462

EN

In the paper thermo-mechanical analysis of Inconel 706 tube welding process was presented. Tubes were joined using electron beam welding EBW. Process simulation was performed using finite element method, FEM Key aspect of welding process simulation is definition of heat source. Geometry of heat source and heat input have direct impact on fusion zone, FZ, and heat affected zone, HAZ. The goal of the work was to design EBW that will produce FZ of required depth. The set of process parameters was identified based on work of Ferro for Inconel 706. Modification of the process parameters was required. For this purpose partial least square method, PLS, was used. PLS model was built using results of own work on EBW for 18-8 steel. The model was applied to Inconel data. The results calculated by PLS model were used to build FEM model.

PL

W pracy przedstawiono analizę termo-mechaniczną procesu spawania tulei wykonanych ze stopu Inconel 706. Tuleje połączono za pomocą wiązki elektronów. Symulacja procesu została wykonana przy użyciu metody elementów skończonych, MES. Kluczowym aspektem przy symulacji procesu spawania jest zdefiniowanie źródła ciepła. Geometria źródła ciepła i ilość wydzielonej mocy mają bezpośredni wpływ na strefę przetopienia i strefę wpływu ciepła. Celem pracy jest zaprojektowanie procesu spawania wiązką elektronów w celu uzyskania określonej głębokości spoiny. Zestaw parametrów procesu spawania został określony na podstawie pracy Ferro wykonanej dla stopu Inconel 706. Wymagana była modyfikacja parametrów przedstawionych w pracy. W tym celu wykorzystano metodę częściowo najmniejszych kwadratów, PLS. Model PLS został zbudowany na podstawie wyników badań własnych dla spawania stali 18-8 za pomocą wiązki elektronów. Opracowany model został wykorzystany do modyfikacji parametrów spawania. Otrzymane wyniki posłużyły do opracowania modelu MES procesu spawania.

8

The use of chemometrics to analyse protein patterns from gel electrophoresis

Faergestad E. M, Rye M. B, Nhek S, Hollung K, Grove H

Acta Chromatographica

|

2011

|

Vol. 23, no. 1

1--40

EN

Chemometrics involves strategies to analyse multivariate data using interdisciplinary approaches aiming to extract relevant information from complex data. Chemometric strategies comprise both the pre-processing of the data, where the choice of methodology is domain-specific, and analysis of the resulting data after preprocessing using multivariate methodology. Although use of multivariate data analysis for gel electrophoresis images has increased substantially in the last decade, its use is still much less frequent than use of univariate approaches. Considering the complexity of the electrophoresis gel images and the multivariate nature of the proteome, applying multivariate data analysis for gel electrophoresis images gives information which is otherwise lost. This paper is written as a review and guideline of chemometric strategies used for analysis of gel electrophoresis images. The multivariate data analyses described are, however, also relevant for other proteome data, for example mass spectrometry, and for functional genomics in general.

9

Stability of gene selection methods for multiclass clssification

Student S., Fujarewicz K.

Journal of Medical Informatics & Technologies

|

2010

|

Vol. 15

101--107

EN

A big problem in applying DNA microarrays for classification is dimension of the dataset. Recently we proposed a gene selection method based on Partial Least Squares (PLS) for searching best genes for classification. The new idea is to use PLS not only as multiclass approach, but to construct more binary selections that use one versus rest and one versus one approaches. Ranked gene lists are highly instable in the sense, that a small change of the data set often leads to big change of the obtained ordered list. In this article, we take a look at the assessment of stability of our approaches. We compare the variability of the obtained ordered lists from proposed methods with well known Recursive Feature Elimination (RFE) method and classical t-test method. This paper focuses on effective identification of informative genes. As a result, a new strategy to find small subset of significant genes is designed. Our results on real cancer data show that our approach has very high accuracy rate for different combinations of classification methods giving in the same time very stable feature rankings.

10

New Modelling and Process Optimisation Approach for the False-Twist Texturing of Polyester

Silva E. A., Paiva A. P., Balestrassi P. P., Silva C. E. S.

Fibres & Textiles in Eastern Europe

|

2009

|

Vol. 17, no. 6 (77)

57--62

EN

After the wave of ISO 9000 certification, a large number of enterprises started to accumulate a great amount of data regarding their processes. False-twist texturing plants used these data to set up a process and improve their operations. This article shows that data mining, partial least squares modelling and genetic algorithm optimization can provide further use for these data to benefit the company in many areas, such as setting up adequate process parameters without requiring an expert to do so, providing the customer with the requirements that will fulfill his needs, simplifying machine changes, and reducing lot changes. The results show that the model and optimization structure put together can find multiple solutions for machine parameters by providing the multiple product properties or quality levels desired. The prediction of yarn properties, such as linear density (Dtex), elongation, tenacity and boiled water shrinkage were made with R2 between 0.80 and 0.99.

PL

Wprowadzanie norm ISO spowodowało, że duża ilość firm zaczęła gromadzić obszerne dane dotyczące stosowanych procesów produkcyjnych. Przedsiębiorstwa realizujące procesy teksturowania fałszywym skrętem wykorzystywały zgromadzone dane dla ustalenia warunków procesu i udoskonalenia jego przebiegu. W artykule wykazano, że analiza danych, odpowiednie modelowanie oraz optymalizacja z wykorzystaniem algorytmów genetycznych może prowadzić do dalszych udoskonaleń procesu. Wynikiem tego mogą być rozliczne korzyści przedsiębiorstwa polegające na możliwości ustawienia odpowiednich parametrów procesu technologicznego bez potrzeby przeprowadzania dodatkowych doświadczeń oraz udostępnienia klientowi zestawu parametrów spełniających jego wymagania. Dzięki temu można również uzyskać uproszczenie wymiany maszyn oraz zmniejszenie ilość partii próbnych. Wyniki wykazały, że model opracowany wspólnie ze strukturą optymalizacji może doprowadzić do znalezienia wielu zestawów parametrów potrzebnych dla uzyskania określonych asortymentów produkcji oraz wymaganej jakości. Przewidywanie takich parametrów włókien jak gęstość liniowa, wytrzymałość liniowa, wydłużenie przy zerwaniu i skurcz we wrzącej wodzie można było określić przy współczynniku R2 w granicach od 0.80 do 0.99.

11

Maximum Separation Partial Least Squares (MSPLS): a new method for classification in microarray experiment

Błaszczyk P., Stąpor K.

Journal of Medical Informatics & Technologies

|

2007

|

Vol. 11

187--195

EN

The purpose of the paper is to propose a new method for classification. Our MSPLS method was deduced from the classic Partial Least Squares (PLS) algorithm. In this method we applied the Maximum Separation Criterion. On the basis of the approach we are able to find such weight vectors that the dispersion between the classes is maximal and the dispersion within the classes is minimal. In order to compare the performance of classifier we used the following types of dataset - biological and simulated. Error rates and confidence intervals were estimated by the jackknife method.