Preferencje help
Widoczny [Schowaj] Abstrakt
Liczba wyników

Znaleziono wyników: 16

Liczba wyników na stronie
first rewind previous Strona / 1 next fast forward last
Wyniki wyszukiwania
Wyszukiwano:
w słowach kluczowych:  boosting
help Sortuj według:

help Ogranicz wyniki do:
first rewind previous Strona / 1 next fast forward last
EN
The paper presents new ensemble solutions, which can forecast the average level of particulate matters PM10 and PM2.5 with increased accuracy. The proposed network is composed of weak predictors integrated into a final expert system. The members of the ensemble are built based on deep multilayer perceptron and decision tree and use bagging and boosting principle in elaborating common decisions. The numerical experiments have been carried out for prediction of daily average pollution of PM10 and PM2.5 for the next day. The results of experiments have shown, that bagging and boosting ensembles employing these weak predictors improve greatly the quality of results. The mean absolute errors have been reduced by more than 30% in the case of PM10 and 20% in the case of PM2.5 in comparison to individually acting predictors.
EN
This paper describes a number of experiments to compare and validate the performance of machine learning classifiers. Creating machine learning models for data with wide varieties has huge applications in predictive modelling across multiple domain of science. This work reviews state of the art techniques in machine learning classifiers methods with several extent of magnitude in statistics and key findings that will be helpful in establishing best methodological practices for class predictions. Comprehensive comparative review analysis with statistical validations for various machine learning algorithm for SVM, Bagging, Boosting, Decision Trees and Nearest Neighborhood algorithm on multiple data sets is carried out. Focus on the statistical analysis of the results using Friedman-Test and Wilcoxon Test as well as other interpretative metrics like classification rate, ROC, F-measure are evaluated to benchmark results.
EN
Supervised classification covers a number of data mining methods based on training data. These methods have been successfully applied to solve multi-criteria complex classification problems in many domains, including economical issues. In this paper we discuss features of some supervised classification methods based on decision trees and apply them to the direct marketing campaigns data of a Portuguese banking institution. We discuss and compare the following classification methods: decision trees, bagging, boosting, and random forests. A classification problem in our approach is defined in a scenario where a bank’s clients make decisions about the activation of their deposits. The obtained results are used for evaluating the effectiveness of the classification rules.
EN
Downsizing is one of the development trends of internal combustion engine due to its direct impact on fuel economy and indirectly in reducing the emission of carbon dioxide into the atmosphere. Changing the displacement associated with the same engine performance needs support by additional systems, which primarily include the boost. This paper describes downsizing idea, a review of recharging methods and thermodynamic analysis of the combustion process for the chosen engine before and after downsizing taking into account the different variants of boost. The core objective of this study is to downsize a naturally aspirated 1.6L BMW PSA engine by 25% of its initial swept volume and then boosting downsized engine with higher-pressure ratio using the turbocharger set. The study focuses on the analysis of four turbochargers from Garrett turbos. The study winds up with the analysis of engine performance based on the values of compression ratio, air-fuel ratio, polytropic exponents of compression and decompression with keeping the same chemical composition of the fuel. At the end, study was resulted with turbocharger Garrett GT1548 as a the best solution form considered ones, because of: wide range of pressure rate, reasonably sufficient for the engine of this size, enough room (60%) for extracting better performance, lower compression ratio value, which counts the rise of brake mean effective pressure, although to a very little extent and leaner mixture at 1,20 value of the air/fuel ratio with maximum power and reduction of fuel consumption, what was satisfied for downsizing techniques.
EN
The main disadvantages of two-stroke engines such a big fuel consumption and big emission of hydrocarbons or carbon monoxide can be reduced by new proposal of design of two stroke engine based on four stroke engines. The paper describes the operation of high supercharged spark ignition overhead poppet valve two-stroke engine, which enables to achieve higher total efficiency and exhaust gas emission comparable to four-stroke engines. The work of such engines is possible by proper choice of valve timings, geometrical parameters of inlet and outlet ducts and charge pressure. The engine has to be equipped with direct fuel injection system enabling lower emission of pollutants. The work is based on theoretical considerations and engine parameters are determined on the simulation process by use GT-Power program and CFD program for different engine configurations. The initial results included in the paper show influence of valve timing on engine work parameters and predicted exhaust gas emission. The simulation results show that the nitrogen oxides are considerably reduced in comparison to four-stroke engines because of higher internal exhaust gas recirculation. The innovation of this proposal is applying of variable valve timing with turbocharging system in the two-stroke engine and obtaining a significant downsizing effect. The conclusions shows the possibilities of applying two-stroke poppet valve engine as a power unit for transportation means with higher total efficiency than traditional engines with possible change of engine operation in two modes: two- and four stroke cycles.
PL
Główne wady silników dwusuwowych, takie jak duże zużycie paliwa i duża emisja węglowodorów oraz tlenku węgla mogą być wyeliminowane przez propozycję nowej konstrukcji silnika 2-suwowego opartego na silniku 4-suwowym. W artykule przedstawiono pracę wysoko doładowanego górnozaworowego silnika 2-suwowego, który umożliwia uzyskanie większej sprawności niż silnika 4-suwowego przy utrzymaniu emisji szkodliwych składników spalin na poziomie silników 4-suwowych. Silnik jest wyposażony w bezpośredni wtrysk paliwa umożliwiający mniejszą emisję szkodliwych składników spalin. Przedstawiono osiągi silnika, które określono na podstawie symulacji w programie GT-Power oraz przy użyciu programu CFD. Silnik został zmodyfikowany przez zmianę systemu zaworowego i faz rozrządu dla pracy w cyklu 2-suwowym. Zaprezentowano zmiany zachodzące w parametrach termodynamicznych cylindra w czasie wymiany ładunku dla wersji 2- i 4-suwowej. Proces spalania był modelowany za pomocą prostych reakcji kinetycznych dla węglowodorów i tlenków azotu oraz reakcji dysocjacji. Wyniki symulacji wskazują, że emisja tlenków azotu jest znacznie mniejsza w porównaniu z silnikiem 4-suwowym z powodu dużej wewnętrznej recyrkulacji spalin. Innowacją w proponowanym silniku dwusuwowym jest zmienność faz rozrządu z doładowaniem turbosprężarkowym i uzyskanie znacznego zmniejszenia gabarytów silnika (downsizing). Podano także możliwość zastosowania górnozaworowego silnika 2-suwowego jako jednostki napędowej dla środków transportu o większej sprawności niż tradycyjne silniki, z możliwością pracy zarówno w cyklu 2-suwowym, jak i 4-suwowym.
EN
Boosting is a classification method which has been proven useful in non-satellite image processing while it is still new to satellite remote sensing. It is a meta-algorithm, which builds a strong classifier from many weak ones in iterative way. We adapt the AdaBoost.M1 boosting algorithm in a new land cover classification scenario based on utilization of very simple threshold classifiers employing spectral and contextual information. Thresholds for the classifiers are automatically calculated adaptively to data statistics. The proposed method is employed for the exemplary problem of artificial area identification. Classification of IKONOS multispectral data results in short computational time and overall accuracy of 94.4% comparing to 94.0% obtained by using AdaBoost.M1 with trees and 93.8% achieved using Random Forest. The influence of a manipulation of the final threshold of the strong classifier on classification results is reported.
7
Content available remote Boosting, bagging and fixed fusion methods performance for aiding diagnosis
EN
Multiple classifier fusion may generate more accurate classification than each of the constituent classifiers. The aim was to examine the ensemble performance by the comparison of boosting, bagging and fixed fusion methods for aiding diagnosis. Real-life medical data set for thyroid diseases recognition was applied. Different fixed combined classifiers (mean, average, product, minimum, maximum, and majority vote) built on parametric and nonparametric Bayesian discriminant methods have been employed. No very significant improvement of recognition rates by a fixed classifier combination was achieved on the examined data. The best performance was obtained for resampling methods with classification trees, for both the bagging and the boosting combining methods. The bagging and the boosting logistic regression methods have proven less efficient than the bagging or the boosting of neural networks. Difference between the bagging and the boosting performance for the examined data set was not obtained.
8
Content available Gas exchange in valved two-stroke SI engine
EN
The paper describes the work of high speed charged spark ignition overhead poppet valve two-stroke engine, which enables to achieve higher total efficiency and exhaust gas emission comparable to four-stroke engines. The work of such engines is possible by proper choice ofvalve timings, geometrical parameters of inlet, outlet ducts and charge pressure. The engine has to be equipped with direct fuel injection system enabling lower emission of pollutants. The work is based on theoretical considerations performed in GT-Power in previous authors' research and carried out in CFD code (KIVA 3 V) for different engine configurations. The initial results included in the paper show influence of inlet port geometry and charge pressure on engine scavenging process. Additionally, optimum fuel spray injector position was considered in order to obtain proper fuel vaporization and avoid significant wall-wetting. The simulation results show that the nitrogen oxides arę considerably reduced in comparison to four-stroke engines because ofhigher internal exhaust gas recirculation. The innovation of this proposal is applying of poppet intake and exhaust valves with turbocharging in the two-stroke engine and obtaining a significant downsizing effect. The conclusion shows the possibilities of proper gas exchange process in this type of two-stroke engine and thus, the feasibility of its application as a power unii for transportation means with higher total efficiency than traditional engines with possible change of engine work in two modes: two- and four-stroke cycles.
9
Content available remote Evolving ensembles of linear classifiers by means of clonal selection algorithm
EN
Artificial immune systems (AIS) have become popular among researchers and have been applied to a variety of tasks. Developing supervised learning algorithms based on metaphors from the immune system is still an area in which there is much to explore. In this paper a novel supervised immune algorithm based on clonal selection framework is proposed. It evolves a population of linear classifiers used to construct a set of classification rules. Aggregating strategies, such as bagging and boosting, are shown to work well with the proposed algorithm as the base classifier.
EN
We focus on the adaptation of boosting to representation spaces composed of different subsets of features. Rather than imposing a single weak learner to handle data that could come from different sources (e.g., images and texts and sounds), we suggest the decomposition of the learning task into several dependent sub-problems of boosting, treated by different weak learners, that will optimally collaborate during the weight update stage. To achieve this task, we introduce a new weighting scheme for which we provide theoretical results. Experiments are carried out and show that ourmethod works significantly better than any combination of independent boosting procedures.
11
Content available remote Możliwości obniżenia temperatury topienia szkła opakowaniowego
PL
Składy chemiczne szkieł opakowaniowych od 1932 r. pomimo zawsze istniejących cech podobieństwa były systematycznie unowocześniane, dostosowywane do warunków ekonomicznych i wymagań użytkowników. Do najważnieszych zmian należy zaliczyć: ograniczenie alkalii i zwiększenie CaO, wprowadzenie małych dodatków MgO za CaO, restrykcyjne ograniczenie metali ciężkich, wprowadzenie małych dodatków Li2O, głównie w celu podwyższenia wydajności topienia i poprawienia jakości masy szklanej. W ostatnich latach, z powodu coraz wyższych cen energii cieplnej, zwiększających się wymagań ochrony środowiska i coraz większego zainteresowania się globalnym zjawiskiem cieplarnianym, prowadzone są prace nad zmianami składu chemicznego szkła w celu znacznego obniżenia temperatury topienia, umożliwiającego obniżenie zużycia paliwa, a tym samym zmniejszenie emisji CO2 i NOx. Poprzez zastosowanie tzw. chemicznego ,,boostingu'' tj. wprowadzenie do składu chemicznego szkła synergicznie działających tlenków obniżających lepkość wysoko-temperaturową można obniżyć temperaturę topienia nawet około 100 stopni Celsjusza. Największy wpływ na obniżenie temperatury log eta=2 ma tlenek litowy. Wprowadzenie 0,135% Li2O do składu szkła opakowaniowego obniża temperaturę topienia o około 11 stopni Celsjusza. Przeprojektowanie składu chemicznego szkła musi być dokonane w taki sposób aby zachować należytą jakość szkła, właściwości fizyko-chemiczne, a jednocześnie zachować (z możliwie najmniejszymi zmianami) dotychczasowe warunki formowania, z których najważniejsze to czas stygnięcia i temperatura likwidusu. Kompleksowe spełnienie tych warunków jest jednak skomplikowane.
EN
Future stringent legislation on emissions in combination with the market request of an increase in engine efficiency and optimization poses a great challenge to the engine and components manufacturers. The technologies developed in the last years for Spark Ignition (SI) engines such as turbocharging and variable valve actuation are not able to totally satisfy the future normative. More progress still has to be made in terms of in-cylinder combustion process and efficiency. The aim of this paper is the optimisation of a boosted SI engine in terms of performances, fuel consumption and pollutants emissions with low costs. The experimental activity was carried out on a port fuel injection SI optical engine, equipped with a commercial four-valve head. Innovative injection strategies were tested: in particular, single and double injections were performed when the intake valves were open. Optical techniques based on 2D-digital imaging were used to follow the fuel injection in the intake manifold and simultaneously the flame propagation in the combustion chamber. Conventional measurements of engine parameters and exhaust emissions completed the experimental investigations. The tests demonstrated that the double injection strategies were characterized by higher combustion process efficiency than single injection on. The injection splitting resulted a suitable solution for the reduction in pollutants concentration in the combustion chamber and at the exhaust with a good compromise between performance and fuel consumption.
EN
In this paper, low-cost solutions were proposed to reduce the fuel consumption in a boosted port fuel injection spark ignition (PFI SI) engine, taking into account the engine performances and the pollutants emission. To this purpose, the optical characterization of the fuel injection and of the combustion process was carried out in a PFI SI engine. The experiments were performed on a partially transparent single-cylinder SI engine, equipped with a four-valve head and an external boost device. The intake manifold was optically accessible through three holes that allowed the introduction of an endoscope and of optical fibres. The standard injection condition planned by the engine manufacturer was investigated; it consisted in the fuel injection at 3.5 bar when the intake valves were closed. Moreover, the fuel injection with open intake valves was tested; 3.5 and 6.5 bar fuel pressures were studied for open and closed valves conditions. Optical techniques based on 2D-digital imaging were used to follow the fuel injection spray in the intake manifold and the flame propagation in the combustion chamber. The results of in-cylinder optical investigations were correlated with the engine performances and with the exhaust emissions.
14
Content available remote Weighted ensemble boosting for robust activity recognition in video
EN
In this paper we introduce a novel approach to classifier combination, which we term Weighted Ensemble Boosting. We apply the proposed algorithm to the problem of activity recognition in video, and compare its performance to different classifier combination methods. These include Approximate Bayesian Combination, Boosting, Feature Stacking, and the more traditional Sum and Product rules. Our proposed Weighted Ensemble Boosting algorithm combines the Bayesian averaging strategy with the boosting framework, finding useful conjunctive feature combinations and achieving a lower error rate than the traditional boosting algorithm. The method demonstrates a comparable level of stability with respect to the classifier selection pool. We show the performance of our technique for a set of 6 types of classifiers in an office setting, detecting 7 classes of typical office activities.
15
Content available remote Training set size in ensemble feature selection for clinical proteomics
PL
Spektrometria masowa typu SELDI-TOF została w ostatnich latach zastosowana do diagnostyki nowotworów. W tym celu wykorzystywane są próbki płynów ustrojowych, które poddawane są analizie proteomicznej z wykorzystaniem spektrometru masowego. W efekcie uzyskiwany jest wielowymiarowy opis pobranej próbki. Dla zbioru pacjentów z oraz bez nowotworu możliwe jest stworzenie metody diagnostycznej opartej na metodach uczenia maszynowego. W niniejszym artykule analizujemy efekt wielkości zbioru trenującego używanego do uczenia klasyfikatora rozróżniającego próbki krwi od pacjentów zdrowych i z obecnym nowotworem jajnika. Użyty klasyfikator bazuje na metodzie zespołowej typu boosting używającej reguły decyzyjnej Fishera. Klasyfikator ten został rozszerzony o metodę selekcji cech. W artykule wykazano, iż dla odpowiedniego typu mikromacierzy białkowej użytej w metodzie spektrometrycznej SELDI-TOF, zbiór treningowy zawierający ok. 30-40 próbek pozwala na stworzenie klasyfikatora wykazującego 95%-ową dokładność klasyfikacji. Zintegrowana z klasyfikatorem metoda selekcji cech pozwala na skuteczną klasyfikację przy użyciu tylko 2 cech z widma spektrometrycznego.
EN
The SELDI-TOF mass spectrometry has been recently shown to be useful in diagnosis of a range of cancer types. In the procedure, samples of body fluids are subject to proteomic analysis using mass spectrometry, resulting in highly dimensional fingerprints. The fingerprints gathered from a set of cancer and control patients allow for creation of a machine learning-based method for diagnosing cancer. In this paper, we analyse the effects of the number of examples in the training set used for constructing a classifier distinguishing blood samples from normal and ovarian cancer patients. We employ a version of our FLD boosting classifier, extended to include a feature selection algorithm within a single machine-learning framework. We show that when a particular type of protein chip is used in SELDI-TOF-MS analysis, the training set containing samples from only ca. 30-40 patients is suitable for creating a classifier exhibiting ca. 95% accuracy, sensitivity and specificity to ovarian cancer. The feature selection procedure incorporated into the classifier reduces to 2 the number of mass/charge values used for discrimination.
first rewind previous Strona / 1 next fast forward last
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.