Wyniki wyszukiwania - BazTech

1

Extracting mass concentration time series features for classifcation of indoor and outdoor atmospheric particulates

Hussain Lal, Aziz Wajid, Saeed Sharjil, Rafque Muhammad, Nadeem Malik Sajjad Ahmed, Shim Seong-O, Aftar Sania, Pirzada Jawad-ur-Rehman

Acta Geophysica

|

2020

|

Vol. 68, no. 3

945--963

EN

Particulate matters (PMs) are considered as one of the air pollutants generally associated with poor air quality in both outdoor and indoor environments. The composition, distribution and size of these particles hazardously afect the human health causing cardiovascular health problems, lung dysfunction, respiratory problems, chronic obstructive pulmonary disease and lungs cancer. Classifcation models developed by analyzing mass concentration time series data of atmospheric particulate matter can be used for the prediction of air quality and for issuing warnings to protect the health of the public. In this study, mass concentration time series data of both outdoor and indoor particulates matters PM2.5 (aerodynamics size up to 2.5 μ) and PM10.0 (aerodynamics size up to 10.0 μ) were acquired using Haz-Dust EPAM-5000 from six diferent locations of the Muzafarabad city, Azad Kashmir. The linear and nonlinear approaches were used to extract mass concentration time series features of the indoor and outdoor atmospheric particulates. These features were given as an input to the robust machine learning classifers. The support vector machine (SVM) kernels, ensemble classifers, decision tree and K-nearest neighbors (KNN) are used to classify the indoor and outdoor particulate matter time series. The performance was estimated in terms of area under the curve (AUC), accuracy, true negative rate, true positive rate, negative predictive value and positive predictive value. The highest accuracy (95.8%) was obtained using cubic and coarse Gaussian SVM along with the cosine and cubic KNN, while the highest AUC, i.e., 1.00, is obtained using fne Gaussian and cubic SVM as well as with the cubic and weighted KNN.

2

Polish emotional speech recognition based on the committee of classifiers

Kamińska D., Sapiński T.

Przegląd Elektrotechniczny

|

2017

|

R. 93, nr 6

101--105

EN

This article presents the novel method for emotion recognition from polish speech. We compared two different databases: spontaneous and acted out speech. For the purpose of this research we gathered a set of audio samples with emotional information, which serve as input database. Multiple Classifier Systems were used for classification, with commonly used speech descriptors and different groups of perceptual coefficients as features extracted from audio samples.

PL

Niniejsza praca dotyczy rozpoznawania stanów emocjonalnych na podstawie głosu. W artykule porównaliśmy mowę spontaniczną z mową odegraną. Na potrzeby zrealizowanych badań zgromadzone zostały emocjonalne nagrania audio, stanowiące kompleksową bazę wejściową. Przedstawiamy nowatorski sposób klasyfikacji emocji wykorzystujący komitety klasyfikujące, stosując do opisu emocji powszechnie używane deskryptory sygnału mowy oraz percepcyjne współczynniki hybrydowe.

3

Offline handwritten pre-segmented character recognition of Gurmukhi script

Kumar M., Jindal M. K., Sharma R. K., Jindal S. J.

Machine Graphics and Vision

|

2016

|

Vol. 25, No. 1/4

45--55

EN

In this paper, we have proposed a feature extraction technique for recognition of segmented handwritten characters of Gurmukhi script. The experiments have been performed with 7000 specimens of segmented offline handwritten Gurmukhi characters collected from 200 different writers. We have considered the set of 35 basic characters of the Gurmukhi script and have proposed the feature extraction technique based on boundary extents of the character image. PCA based feature selection technique has also been implemented in this work to reduce the dimension of data. We have used k-NN, SVM and MLP classifiers. SVM has been used with four different kernels. In this work, we have achieved maximum recognition accuracy of 93.8% for the 35-class problem when SVM with RBF kernel and 5-fold cross validation technique were employed.

4

Classification algorithms to identify changes in resistance

Duraj A., Korzeniewska E., Krawczyk A.

Przegląd Elektrotechniczny

|

2015

|

R. 91, nr 12

80-82

EN

In the article the basic method for measuring the resistance of medical electrodes, made based on a thin conductive layer formed during the PVD process, is described. The authors also briefly characterized two algorithms for data classification: k-nearest neighbors and Bayes classifier, which were used as algorithms to detect changes in the electrode resistance.

PL

W artykule została opisana podstawowa metoda pomiaru rezystancji elektrod medycznych wykonanych w oparciu o cienkie warstwy przewodzące powstałe w procesie PVD. Scharakteryzowano również krótko dwa algorytmy klasyfikacji danych: algorytm k najbliższych sąsiadów oraz klasyfikator bayowski, które zostały wykorzystane jako algorytmy identyfikacji zmian rezystancji elektrod.

5

A k-Nearest Neighbors Method for Classifying User Sessions in E-Commerce Scenario

Suchacka G., Skolimowska-Kulig M., Potempa A.

Journal of Telecommunications and Information Technology

|

2015

|

nr 3

64--69

EN

This paper addresses the problem of classification of user sessions in an online store into two classes: buying sessions (during which a purchase confirmation occurs) and browsing sessions. As interactions connected with a purchase confirmation are typically completed at the end of user sessions, some information describing active sessions may be observed and used to assess the probability of making a purchase. The authors formulate the problem of predicting buying sessions in a Web store as a supervised classification problem where there are two target classes, connected with the fact of finalizing a purchase transaction in session or not, and a feature vector containing some variables describing user sessions. The presented approach uses the k-Nearest Neighbors (k-NN) classification. Based on historical data obtained from online bookstore log files a k-NN classifier was built and its efficiency was verified for different neighborhood sizes. A 11-NN classifier was the most effective both in terms of buying session predictions and overall predictions, achieving sensitivity of 87.5% and accuracy of 99.85%.

6

Music genre classification applied to bass enhancement for mobile technology

Hoffmann P., Kostek B.

Elektronika : konstrukcje, technologie, zastosowania

|

2015

|

Vol. 56, nr 4

14-19

EN

The aim of this paper is to present a novel approach to the Virtual Bass Synthesis (VBS) algorithms applied to portable computers. The proposed algorithm is related to intelligent, rule-based setting of synthesis parameters according to music genre of an audio excerpt. The classification of music genres is automatically executed employing MPEG 7 parameters and the Principal Component Analysis method applied to reduce information redundancy. To perform genre recognition k-Nearest Neighbors classifier is used. The VBS algorithm is based on nonlinear device (NLD) or phase vocoder (PV) depending on the content of an audio file excerpt. A soft computing (fuzzy logic) algorithm is employed to set optimum synthesis parameters depending on a given song. To confirm the relationship between genres and preferences of listeners in the low frequency range the pair wise subjective comparison test is carried out. In tests 30 pairs of audio files are employed divided into six popular musical genres. Music excerpts processed by a commercially available bass boost algorithm are used for comparison. Based on the responses of the listeners the statistical analysis is carried out. A short summary is also provided that contains plans for future algorithm development.

PL

Artykuł przedstawia nową metodę wirtualnej syntezy niskich częstotliwości w urządzeniach mobilnych. Proponowana metoda generuje inteligentne reguły do określenia optymalnych parametrów syntezy w zależności od słuchanego gatunku muzycznego. Klasyfikacja gatunków muzycznych odbywa się automatycznie bazując na wektorze parametrów MPEG 7 oraz analizie głównych składowych (ang. Principal Component Analysis) w celu redukcji nadmiarowości informacji. W rozpoznawaniu gatunku muzycznego wykorzystywany jest klasyfikator k-Najbliższych Sąsiadów. Synteza niskich częstotliwości opiera się na metodzie elementów nieliniowych (NLD) lub wokodera fazowego (PV) w zależności od fragmentu utworu. W celu uzyskania optymalnych parametrów syntezy w zależności od słuchanego utworu stosuje się przygotowane reguły logiki rozmytej. Dla potwierdzenia zależności pomiędzy gatunkami muzycznymi a preferencjami słuchaczy w zakresie niskich częstotliwości przeprowadzono testy subiektywne porównań parami. W teście zostało wykorzystanych 30 par utworów należących do sześciu popularnych gatunków muzycznych. Jako odniesienie w teście zostały wykorzystane próbki przetworzone przez komercyjny algorytm należący do grupy Bass Boost. Na podstawie odpowiedzi słuchaczy, uzyskanych w testach subiektywnych, przeprowadzono analizę statystyczną, która potwierdza potrzebę rozpoznawania gatunku muzycznego w kontekście poprawy brzmienia niskich częstotliwości.