To see the other types of publications on this topic, follow the link: Feature Extraction and Classification.
Dissertations / Theses on the topic 'Feature Extraction and Classification'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Feature Extraction and Classification.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
1
Liu, Raymond. "Feature extraction in classification." Thesis, Imperial College London, 2013. http://hdl.handle.net/10044/1/23634.
Full textAbstract:
Feature extraction, or dimensionality reduction, is an essential part of many machine learning applications. The necessity for feature extraction stems from the curse of dimensionality and the high computational cost of manipulating high-dimensional data. In this thesis we focus on feature extraction for classification. There are several approaches, and we will focus on two such: the increasingly popular information-theoretic approach, and the classical distance-based, or variance-based approach. Current algorithms for information-theoretic feature extraction are usually iterative. In contrast, PCA and LDA are popular examples of feature extraction techniques that can be solved by eigendecomposition, and do not require an iterative procedure. We study the behaviour of an example of iterative algorithm that maximises Kapur's quadratic mutual information by gradient ascent, and propose a new estimate of mutual information that can be maximised by closed-form eigendecomposition. This new technique is more computationally efficient than iterative algorithms, and its behaviour is more reliable and predictable than gradient ascent. Using a general framework of eigendecomposition-based feature extraction, we show a connection between information-theoretic and distance-based feature extraction. Using the distance-based approach, we study the effects of high input dimensionality and over-fitting on feature extraction, and propose a family of eigendecomposition-based algorithms that can solve this problem. We investigate the relationship between class-discrimination and over-fitting, and show why the advantages of information-theoretic feature extraction become less relevant in high-dimensional spaces.
2
Goodman, Steve. "Feature extraction and classification." Thesis, University of Sunderland, 2000. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.301872.
Full text
3
Elliott, Rodney Bruce. "Feature extraction techniques for grasp classification." Thesis, University of Canterbury. Mechanical Engineering, 1998. http://hdl.handle.net/10092/3447.
Full textAbstract:
This thesis examines the ability of four signal parameterisation techniques to provide discriminatory information between six different classes of signal. This was done with a view to assessing the suitability of the four techniques for inclusion in the real-time control scheme of a next generation robotic prosthesis. Each class of signal correlates to a particular type of grasp that the robotic prosthesis is able to form. Discrimination between the six classes of signal was done on the basis of parameters extracted from four channels of electromyographie (EMG) data that was recorded from muscles in the forearm. Human skeletal muscle tissue produces EMG signals whenever it contracts. Therefore, providing that the EMG signals of the muscles controlling the movements of the hand vary sufficiently when forming the different grasp types, discrimination between the grasps is possible. While it is envisioned that the chosen command discrimination system will be used by mid-forearm amputees to control a robotic prosthesis, the viability of the different parameterisation techniques was tested on data gathered from able-bodied volunteers in order to establish an upper limit of performance. The muscles from which signals were recorded are: the extensor pollicis brevis and extensor pollicis longus pair (responsible for moving the thumb); the extensor communis digitorum (responsible for moving the middle and index fingers); and the extensor carpi ulnaris (responsible for moving the little finger). The four signal parameterisation techniques that were evaluated are: 1. Envelope Maxima. This method parameterises each EMG signal by the maximum value of a smoothed fitted signal envelope. A tenth order polynomial is fitted to the rectified EMG signal peaks, and the maximum value of the polynomial is used to parameterise the signal. 2. Orthogonal Decomposition. This method uses a set of orthogonal functions to decompose the EMG signal into a finite set of orthogonal components. Each burst is then parameterised by the coefficients of the set of orthogonal functions. Two sets of orthogonal functions were tested: the Legendre polynomials, and the wavelet packets associated with the scaling functions of the Haar wavelet (referred to as the Haar wavelet for brevity). 3. Global Dynamical Model. This method uses a discretised set of nonlinear ordinary differential equations to model the dynamical processes that produced the recorded EMG signals. The coefficients of this model are then used to parameterise the EMG signal 4. EMG Histogram. This method formulates a histogram detailing the frequency with which the EMG signal enters particular voltage bins) and uses these frequency measurements to parameterise the signal. Ten sets of EMG data were gathered and processed to extract the desired parameters. Each data set consisted of 600 grasps- lOO grasp records of four channels of EMG data for each of the six grasp classes. From this data a hit rate statistic was formed for each feature extraction technique. The mean hit rates obtained from the four signal parameterisation techniques that were tested are summarised in Table 1. The EMG histogram provided Parameterisation Technique Hit Rate (%) Envelope Maxima 75 Legendre Polynomials 77 Haar Wavelets 79 Global Dynamical Model 75 EMG Histogram 81 Table 1: Hit Rate Summary. the best mean hit rate of all the signal parameterisation techniques of 81%. However, like all of the signal parameterisations that were tested, there was considerable variance in hit rates between the ten sets of data. This has been attributed to the manner in which the electrodes used to record the EMG signals were positioned. By locating the muscles of interest more accurately, consistent hit rates of 95% are well within reach. The fact that the EMG histogram produces the best mean hit rates is surprising given its relative simplicity. However, this simplicity makes the EMG histogram feature ideal for inclusion in a real-time control scheme.
4
Chilo, José. "Feature extraction for low-frequency signal classification /." Stockholm : Fysik, Physics, 2008. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-4661.
Full text
5
Graf, Arnulf B. A. "Classification and feature extraction in man and machine." [S.l. : s.n.], 2004. http://deposit.ddb.de/cgi-bin/dokserv?idn=972533508.
Full text
6
Hamsici, Onur C. "Bayes Optimality in Classification, Feature Extraction and Shape Analysis." The Ohio State University, 2008. http://rave.ohiolink.edu/etdc/view?acc_num=osu1218513562.
Full text
7
Nilsson, Mikael. "On feature extraction and classification in speech and image processing /." Karlskrona : Department of Signal Processing, School of Engineering, Blekinge Institute of Technology, 2007. http://www.bth.se/fou/forskinfo.nsf/allfirst2/fcbe16e84a9ba028c12573920048bce9?OpenDocument.
Full text
8
Coath, Martin. "A computational model of auditory feature extraction and sound classification." Thesis, University of Plymouth, 2005. http://hdl.handle.net/10026.1/1822.
Full textAbstract:
This thesis introduces a computer model that incorporates responses similar to those found in the cochlea, in sub-corticai auditory processing, and in auditory cortex. The principle aim of this work is to show that this can form the basis for a biologically plausible mechanism of auditory stimulus classification. We will show that this classification is robust to stimulus variation and time compression. In addition, the response of the system is shown to support multiple, concurrent, behaviourally relevant classifications of natural stimuli (speech). The model incorporates transient enhancement, an ensemble of spectro - temporal filters, and a simple measure analogous to the idea of visual salience to produce a quasi-static description of the stimulus suitable either for classification with an analogue artificial neural network or, using appropriate rate coding, a classifier based on artificial spiking neurons. We also show that the spectotemporal ensemble can be derived from a limited class of 'formative' stimuli, consistent with a developmental interpretation of ensemble formation. In addition, ensembles chosen on information theoretic grounds consist of filters with relatively simple geometries, which is consistent with reports of responses in mammalian thalamus and auditory cortex. A powerful feature of this approach is that the ensemble response, from which salient auditory events are identified, amounts to stimulus-ensemble driven method of segmentation which respects the envelope of the stimulus, and leads to a quasi-static representation of auditory events which is suitable for spike rate coding. We also present evidence that the encoded auditory events may form the basis of a representation-of-similarity, or second order isomorphism, which implies a representational space that respects similarity relationships between stimuli including novel stimuli.
9
Benn, David E. "Model-based feature extraction and classification for automatic face recognition." Thesis, University of Southampton, 2000. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.324811.
Full text
10
Zheng, Yue Chu. "Feature extraction for chart pattern classification in financial time series." Thesis, University of Macau, 2018. http://umaclib3.umac.mo/record=b3950623.
Full text
11
Bekiroglu, Yasemi. "Nonstationary feature extraction techniques for automatic classification of impact acoustic signals." Thesis, Högskolan Dalarna, Datateknik, 2008. http://urn.kb.se/resolve?urn=urn:nbn:se:du-3592.
Full textAbstract:
Condition monitoring of wooden railway sleepers applications are generallycarried out by visual inspection and if necessary some impact acoustic examination iscarried out intuitively by skilled personnel. In this work, a pattern recognition solutionhas been proposed to automate the process for the achievement of robust results. Thestudy presents a comparison of several pattern recognition techniques together withvarious nonstationary feature extraction techniques for classification of impactacoustic emissions. Pattern classifiers such as multilayer perceptron, learning cectorquantization and gaussian mixture models, are combined with nonstationary featureextraction techniques such as Short Time Fourier Transform, Continuous WaveletTransform, Discrete Wavelet Transform and Wigner-Ville Distribution. Due to thepresence of several different feature extraction and classification technqies, datafusion has been investigated. Data fusion in the current case has mainly beeninvestigated on two levels, feature level and classifier level respectively. Fusion at thefeature level demonstrated best results with an overall accuracy of 82% whencompared to the human operator.
12
Masip, Rodó David. "Face Classification Using Discriminative Features and Classifier Combination." Doctoral thesis, Universitat Autònoma de Barcelona, 2005. http://hdl.handle.net/10803/3051.
Full textAbstract:
A mesura que la tecnologia evoluciona, apareixen noves aplicacions en el mon de la classificació facial. En el reconeixement de patrons, normalment veiem les cares com a punts en un espai de alta dimensionalitat definit pels valors dels seus pixels. Aquesta aproximació pateix diversos problemes: el fenomen de la "la maledicció de la dimensionalitat", la presència d'oclusions parcials o canvis locals en la il·luminació. Tradicionalment, només les característiques internes de les imatges facials s'han utilitzat per a classificar, on normalment es fa una extracció de característiques. Les tècniques d'extracció de característiques permeten reduir la influencia dels problemes mencionats, reduint també el soroll inherent de les imatges naturals alhora que es poden aprendre característiques invariants de les imatges facials. En la primera part d'aquesta tesi presentem alguns mètodes d'extracció de característiques clàssics: Anàlisi de Components Principals (PCA), Anàlisi de Components Independents (ICA), Factorització No Negativa de Matrius (NMF), i l'Anàlisi Discriminant de Fisher (FLD), totes elles fent alguna mena d'assumpció en les dades a classificar. La principal contribució d'aquest treball es una nova família de tècniques d'extracció de característiques usant el algorisme del Adaboost. El nostre mètode no fa cap assumpció en les dades a classificar, i construeix de forma incremental la matriu de projecció tenint en compte els exemples mes difícilsPer altra banda, en la segon apart de la tesi explorem el rol de les característiques externes en el procés de classificació facial, i presentem un nou mètode per extreure un conjunt alineat de característiques a partir de la informació externa que poden ser combinades amb les tècniques clàssiques millorant els resultats globals de classificació.
As technology evolves, new applications dealing with face classification appear. In pattern recognition, faces are usually seen as points in a high dimensional spaces defined by their pixel values. This approach must deal with several problems such as: the curse of dimensionality, the presence of partial occlusions or local changes in the illumination. Traditionally, only the internal features of face images have been used for classification purposes, where usually a feature extraction step is performed. Feature extraction techniques allow to reduce the influence of the problems mentioned, reducing also the noise inherent from natural images and learning invariant characteristics from face images. In the first part of this thesis some internal feature extraction methods are presented: Principal Component Analysis (PCA), Independent Component Analysis (ICA), Non Negative Matrix Factorization (NMF), and Fisher Linear Discriminant Analysis (FLD), all of them making some kind of the assumption on the data to classify. The main contribution of our work is a non parametric feature extraction family of techniques using the Adaboost algorithm. Our method makes no assumptions on the data to classify, and incrementally builds the projection matrix taking into account the most difficult samples.
On the other hand, in the second part of this thesis we also explore the role of external features in face classification purposes, and present a method for extracting an aligned feature set from external face information that can be combined with the classic internal features improving the global performance of the face classification task.
13
Magnusson, Ludvig, and Johan Rovala. "AI Approaches for Classification and Attribute Extraction in Text." Thesis, Linnéuniversitetet, Institutionen för datavetenskap (DV), 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:lnu:diva-67882.
Full textAbstract:
As the amount of data online grows, the urge to use this data for different applications grows as well. Machine learning can be used with the intent to reconstruct and validate the data you are interested in. Although the problem is very domain specific, this report will attempt to shed some light on what we call strategies for classification, which in broad terms mean, a set of steps in a process where the end goal is to have classified some part of the original data. As a result, we hope to introduce clarity into the classification process in detail as well as from a broader perspective. The report will investigate two classification objectives, one of which is dependent on many variables found in the input data and one that is more literal and only dependent on one or two variables. Specifically, the data we will classify are sales-objects. Each sales-object has a text describing the object and a related image. We will attempt to place these sales-objects into the correct product category. We will also try to derive the year of creation and it’s dimensions such as height and width. Different approaches are presented in the aforementioned strategies in order to classify such attributes. The results showed that for broader attributes such as a product category, supervised learning is indeed an appropriate approach, while the same can not be said for narrower attributes, which instead had to rely on entity recognition. Experiments on image analytics in conjunction with supervised learning proved image analytics to be a good addition when requiring a higher precision score.
14
Shang, Changjing. "Principal features based texture classification using artificial neural networks." Thesis, Heriot-Watt University, 1995. http://hdl.handle.net/10399/1323.
Full text
15
Dilger, Samantha Kirsten Nowik. "Pushing the boundaries: feature extraction from the lung improves pulmonary nodule classification." Diss., University of Iowa, 2016. https://ir.uiowa.edu/etd/3071.
Full textAbstract:
Lung cancer is the leading cause of cancer death in the United States. While low-dose computed tomography (CT) screening reduces lung cancer mortality by 20%, 97% of suspicious lesions are found to be benign upon further investigation. Computer-aided diagnosis (CAD) tools can improve the accuracy of CT screening, however, current CAD tools which focus on imaging characteristics of the nodule alone are challenged by the limited data captured in small, early identified nodules. We hypothesize a CAD tool that incorporates quantitative CT features from the surrounding lung parenchyma will improve the ability of a CAD tool to determine the malignancy of a pulmonary nodule over a CAD tool that relies solely on nodule features. Using a higher resolution research cohort and a retrospective clinical cohort, two CAD tools were developed with different intentions. The research-driven CAD tool incorporated nodule, surrounding parenchyma, and global lung measurements. Performance was improved with the inclusion of parenchyma and global features to 95.6%, compared to 90.2% when only nodule features were used. The clinically-oriented CAD tool incorporated nodule and parenchyma features and clinical risk factors and identified several features robust to CT variability, resulting in an accuracy of 71%. This study supports our hypothesis that the inclusion of parenchymal features in the developed CAD tools resulted in improved performance compared to the CAD tool constructed solely with nodule features. Additionally, we identified the optimal amount of lung parenchyma for feature extraction and explored the potential of the CAD tools in a clinical setting.
16
Sernheim, Mikael. "Experimental Study on ClassifierDesign and Text Feature Extraction for Short Text Classification." Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-323214.
Full textAbstract:
Text classification is a wide research field with existing ready-to-use solutions for supervised training of text classifiers. The task of classifying short texts puts dif-ferent demands on the invoked learning system that general text classification does not. This thesis explores this challenge by experimenting on how to design the clas-sification system and what text features granted the best results. In the experimental study, a hierarchical versus a flat design was compared, along with different aspects of text features. The method consisted of training and testing on a dataset of 3.2 million samples in total. The test results were evaluated with the quality measures: precision, recall, F1-score and ROC analysis with a modification to target multi-class classification. The result of the experimental study was: 2-level hierarchical designed classifier gave better results than a flat designed classifier in 11 out of 13 occasions; integer represented terms outperformed TFIDF weighted terms of BOW features; lowercase conversion improved the classification results; bigram and tri-gram BOW features achieved better results than unigram BOW features. The results of the experimental study were used in a case study together with Thingmap, which maps natural language queries with users. The case study showed an improvement over earlier solutions of Thingmap’s system.
17
Fargeas, Aureline. "Classification, feature extraction and prediction of side effects in prostate cancer radiotherapy." Thesis, Rennes 1, 2016. http://www.theses.fr/2016REN1S022/document.
Full textAbstract:
Le cancer de la prostate est l'un des cancers les plus fréquents chez l'homme. L'un des traitements standard est la radiothérapie externe, qui consiste à délivrer un rayonnement d'ionisation à une cible clinique, en l'occurrence la prostate et les vésicules séminales. Les objectifs de la radiothérapie externe sont la délivrance d'une dose d'irradiation maximale à la tumeur tout en épargnant les organes voisins (principalement le rectum et la vessie) pour éviter des complications suite au traitement. Comprendre les relations dose/toxicité est une question centrale pour améliorer la fiabilité du traitement à l'étape de planification inverse. Des modèles prédictifs de toxicité pour le calcul des probabilités de complications des tissus sains (normal tissue complication probability, NTCP) ont été développés afin de prédire les événements de toxicité en utilisant des données dosimétriques. Les principales informations considérées sont les histogrammes dose-volume (HDV), qui fournissent une représentation globale de la distribution de dose en fonction de la dose délivrée par rapport au pourcentage du volume d'organe. Cependant, les modèles actuels présentent certaines limitations car ils ne sont pas totalement optimisés; la plupart d'entre eux ne prennent pas en compte les informations non-dosimétrique (les caractéristiques spécifiques aux patients, à la tumeur et au traitement). De plus, ils ne fournissent aucune compréhension des relations locales entre la dose et l'effet (dose-espace/effet relations) car ils n'exploitent pas l'information riche des distributions de planification de dose 3D. Dans un contexte de prédiction de l'apparition de saignement rectaux suite au traitement du cancer de la prostate par radiothérapie externe, les objectifs de cette thèse sont : i) d'extraire des informations pertinentes à partir de l'HDV et des variables non-dosimétriques, afin d'améliorer les modèles NTCP existants et ii) d'analyser les corrélations spatiales entre la dose locale et les effets secondaires permettant une caractérisation de la distribution de dose 3D à l'échelle de l'organe. Ainsi, les stratégies visant à exploiter les informations provenant de la planification (distributions de dose 3D et HDV) ont été proposées. Tout d'abord, en utilisant l'analyse en composantes indépendantes, un nouveau modèle prédictif de l'apparition de saignements rectaux, combinant d'une manière originale l'information dosimétrique et non-dosimétrique, a été proposé. Deuxièmement, nous avons mis au point de nouvelles approches visant à prendre conjointement profit des distributions de dose de planification 3D permettant de déceler la corrélation subtile entre la dose locale et les effets secondaires pour classer et/ou prédire les patients à risque de souffrir d'un saignement rectal, et d'identifier les régions qui peuvent être à l'origine de cet événement indésirable. Plus précisément, nous avons proposé trois méthodes stochastiques basées sur analyse en composantes principales, l'analyse en composantes indépendantes et la factorisation discriminante en matrices non-négatives, et une méthode déterministe basée sur la décomposition polyadique canonique de tableaux d'ordre 4 contenant la dose planifiée. Les résultats obtenus montrent que nos nouvelles approches présentent de meilleures performances générales que les méthodes prédictives de la littératureProstate cancer is among the most common types of cancer worldwide. One of the standard treatments is external radiotherapy, which involves delivering ionizing radiation to a clinical target, in this instance the prostate and seminal vesicles. The goal of radiotherapy is to achieve a maximal local control while sparing neighboring organs (mainly the rectum and the bladder) to avoid normal tissue complications. Understanding the dose/toxicity relationships is a central question for improving treatment reliability at the inverse planning step. Normal tissue complication probability (NTCP) toxicity prediction models have been developed in order to predict toxicity events using dosimetric data. The main considered information are dose-volume histograms (DVH), which provide an overall representation of dose distribution based on the dose delivered per percentage of organ volume. Nevertheless, current dose-based models display limitations as they are not fully optimized; most of them do not include additional non-dosimetric information (patient, tumor and treatment characteristics). Furthermore, they do not provide any understanding of local relationships between dose and effect (dose-space/effect relationship) as they do not exploit the rich information from the 3D planning dose distributions. In the context of rectal bleeding prediction after prostate cancer external beam radiotherapy, the objectives of this thesis are: i) to extract relevant information from DVH and non-dosimetric variables, in order to improve existing NTCP models and ii) to analyze the spatial correlations between local dose and side effects allowing a characterization of 3D dose distribution at a sub-organ level. Thus, strategies aimed at exploiting the information from the radiotherapy planning (DVH and 3D planned dose distributions) were proposed. Firstly, based on independent component analysis, a new model for rectal bleeding prediction by combining dosimetric and non-dosimetric information in an original manner was proposed. Secondly, we have developed new approaches aimed at jointly taking advantage of the 3D planning dose distributions that may unravel the subtle correlation between local dose and side effects to classify and/or predict patients at risk of suffering from rectal bleeding, and identify regions which may be at the origin of this adverse event. More precisely, we proposed three stochastic methods based on principal component analysis, independent component analysis and discriminant nonnegative matrix factorization, and one deterministic method based on canonical polyadic decomposition of fourth order array containing planned dose. The obtained results show that our new approaches exhibit in general better performances than state-of-the-art predictive methods
18
Paraskevas, Ioannis. "Phase as a feature extraction tool for audio classification and signal localisation." Thesis, University of Surrey, 2005. http://epubs.surrey.ac.uk/843856/.
Full textAbstract:
The aim of this research is to demonstrate the significance of signal phase content in time localization issues in synthetic signals and in the extraction of appropriate features from acoustically similar audio recordings (non-synthetic signals) for audio classification purposes. Published work, relating to audio classification, tends to be1 focused on the discrimination of audio classes that are dissimilar acoustically. Consequently, a wide range of features, extracted from the audio recordings, has been appropriate for the classification task. In this research, the audio classification application involves audio recordings (digitized through the same pre-processing conditions) that are acoustically similar and hence, only a few features are appropriate, due to the similarity amongst the classes. The difficulties in processing the phase spectrum of a signal have probably led previous researchers to avoid its investigation. In this research, the sources of these difficulties are studied and certain methods are employed to overcome them. Subsequently, the phase content of the signal has been found to be useful for various applications. The justification of this, is demonstrated via audio classification (non-synthetic signals) and time localization (synthetic signals) applications. Summarizing, the original contributions, introduced based on this research work, are the 'whitened' Hartley spectrum and its short-time analysis, as well as the use of the Hartley phase cepstrum as a time localization tool and the use of phase related feature vectors for the audio classification application.
19
Brown, Dane. "Investigating combinations of feature extraction and classification for improved image-based multimodal biometric systems at the feature level." Thesis, Rhodes University, 2018. http://hdl.handle.net/10962/63470.
Full textAbstract:
Multimodal biometrics has become a popular means of overcoming the limitations of unimodal biometric systems. However, the rich information particular to the feature level is of a complex nature and leveraging its potential without overfitting a classifier is not well studied. This research investigates feature-classifier combinations on the fingerprint, face, palmprint, and iris modalities to effectively fuse their feature vectors for a complementary result. The effects of different feature-classifier combinations are thus isolated to identify novel or improved algorithms. A new face segmentation algorithm is shown to increase consistency in nominal and extreme scenarios. Moreover, two novel feature extraction techniques demonstrate better adaptation to dynamic lighting conditions, while reducing feature dimensionality to the benefit of classifiers. A comprehensive set of unimodal experiments are carried out to evaluate both verification and identification performance on a variety of datasets using four classifiers, namely Eigen, Fisher, Local Binary Pattern Histogram and linear Support Vector Machine on various feature extraction methods. The recognition performance of the proposed algorithms are shown to outperform the vast majority of related studies, when using the same dataset under the same test conditions. In the unimodal comparisons presented, the proposed approaches outperform existing systems even when given a handicap such as fewer training samples or data with a greater number of classes. A separate comprehensive set of experiments on feature fusion show that combining modality data provides a substantial increase in accuracy, with only a few exceptions that occur when differences in the image data quality of two modalities are substantial. However, when two poor quality datasets are fused, noticeable gains in recognition performance are realized when using the novel feature extraction approach. Finally, feature-fusion guidelines are proposed to provide the necessary insight to leverage the rich information effectively when fusing multiple biometric modalities at the feature level. These guidelines serve as the foundation to better understand and construct biometric systems that are effective in a variety of applications.
20
Ren, Bobby (Bobby B. ). "Calibration, feature extraction and classification of water contaminants using a differential mobility spectrometer." Thesis, Massachusetts Institute of Technology, 2006. http://hdl.handle.net/1721.1/53163.
Full textAbstract:
Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2006.Includes bibliographical references (p. 87-89).
High-Field Asymmetric Waveform Ion Mobility Spectrometry (FAIMS) is a chemical sensor that separates ions in the gaseous phase based on their mobility in high electric fields. A threefold approach was developed for both chemical type classification and concentration classification of water contaminants for FAIMS signals. The three steps in this approach are calibration, feature extraction, and classification. Calibration was carried out to remove baseline fluctation and other variations in FAIMS data sets. Four feature extraction algorithms were used to extract subsets of the signal that had high separation potential between two classes of signals. Finally, support vector machines were used for binary classification. The success of classification was measured both by using separability metrics to evaluate the separability of extracted features, and by the percent of correct classification (Pcc) in each task.
by Bobby Ren.
M.Eng.
21
Hapuarachchi, Pasan. "Feature selection and artifact removal in sleep stage classification." Thesis, University of Waterloo, 2006. http://hdl.handle.net/10012/2879.
Full textAbstract:
The use of Electroencephalograms (EEG) are essential to the analysis of sleep disorders in patients. With the use of electroencephalograms, electro-oculograms (EOG), and electromyograms (EMG), doctors and EEG technician can make conclusions about the sleep patterns of patients. In particular, the classification of the sleep data into various stages, such as NREM I-IV, REM, Awake, is extremely important. The EEG signal itself is highly sensitive to physiological and non-physiological artifacts. Trained human experts can accommodate for these artifacts while they are analyzing the EEG signal. However, if some of these artifacts are removed prior to analysis, their job will be become easier. Furthermore, one of the biggest motivations, of our team's research is the construction of a portable device that can analyze the sleep data as they are being collected. For this task, the sleep data must be analyzed completely automatically in order to make the classifications.
The research presented in this thesis concerns itself with the denoising and the feature selection aspects of the teams' goals. Since humans are able to process artifacts and ignore them prior to classification, an automated system should have the same capabilities or close to them. As such, the denoising step is performed to condition the data prior to any other stages of the sleep stage neoclassicisms. As mentioned before, the denoising step, by itself, is useful to human EEG technicians as well.
The denoising step in this research mainly looks at EOG artifacts and artifacts isolated to a single EEG channel, such as electrode pop artifacts. The first two algorithms uses Wavelets exclusively (BWDA and WDA), while the third algorithm is a mixture of Wavelets and In- dependent Component Analysis (IDA). With the BWDA algorithm, determining consistent thresholds proved to be a difficult task. With the WDA algorithm, the performance was better, since the selection of the thresholds was more straight-forward and since there was more control over defining the duration of the artifacts. The IDA algorithm performed inferior to the WDA algorithm. This could have been due to the small number of measurement channels or the automated sub-classifier used to select the denoised EEG signal from the set of ICA demixed signals.
The feature selection stage is extremely important as it selects the most pertinent features to make a particular classification. Without such a step, the classifier will have to process useless data, which might result in a poorer classification. Furthermore, unnecessary features will take up valuable computer cycles as well. In a portable device, due to battery consumption, wasting computer cycles is not an option. The research presented in this thesis shows the importance of a systematic feature selection step in EEG classification. The feature selection step produced excellent results with a maximum use of just 5 features. During automated classification, this is extremely important as the automated classifier will only have to calculate 5 features for each given epoch.
22
Furuhashi, Takeshi, Tomohiro Yoshikawa, Kanta Tachibana, and Minh Tuan Pham. "Feature Extraction Based on Space Folding Model and Application to Machine Learning." 日本知能情報ファジィ学会, 2010. http://hdl.handle.net/2237/20689.
Full textAbstract:
Session ID: TH-F3-4SCIS & ISIS 2010, Joint 5th International Conference on Soft Computing and Intelligent Systems and 11th International Symposium on Advanced Intelligent Systems. December 8-12, 2010, Okayama Convention Center, Okayama, Japan
23
Zilberman, Eric R. "Autonomous time-frequency cropping and feature-extraction algorithms for classification of LPI radar modulations." Thesis, Monterey, Calif. : Springfield, Va. : Naval Postgraduate School ; Available from National Technical Information Service, 2006. http://library.nps.navy.mil/uhtbin/hyperion/06Jun%5FZilberman.pdf.
Full text
24
Schnur, Steven R. "Identification and classification of OFDM based signals using preamble correlation and cyclostationary feature extraction." Thesis, Monterey, California : Naval Postgraduate School, 2009. http://edocs.nps.edu/npspubs/scholarly/theses/2009/Sep/09Sep%5FSchnur.pdf.
Full textAbstract:
Thesis (M.S. in Electrical Engineering)--Naval Postgraduate School, September 2009.Thesis Advisor(s): Tummala, Murali ; McEachen, John. "September 2009." Description based on title screen as viewed on November 5, 2009. Author(s) subject terms: IEEE 802.11, IEEE 802.16, OFDM, Cyclostationary Feature Extraction, FFT Accumulation Method. Includes bibliographical references (p. 103-104). Also available in print.
25
Smith, R. S. "Angular feature extraction and ensemble classification method for 2D, 2.5D and 3D face recognition." Thesis, University of Surrey, 2008. http://epubs.surrey.ac.uk/843069/.
Full textAbstract:
It has been recognised that, within the context of face recognition, angular separation between centred feature vectors is a useful measure of dissimilarity. In this thesis we explore this observation in more detail and compare and contrast angular separation with the Euclidean, Manhattan and Mahalonobis distance metrics. This is applied to 2D, 2.5D and 3D face images and the investigation is done in conjunction with various feature extraction techniques such as local binary patterns (LBP) and linear discriminant analysis (LDA). We also employ error-correcting output code (ECOC) ensembles of support vector machines (SVMs) to project feature vectors non-linearly into a new and more discriminative feature space. It is shown that, for both face verification and face recognition tasks, angular separation is a more discerning dissimilarity measure than the others. It is also shown that the effect of applying the feature extraction algorithms described above is to considerably sharpen and enhance the ability of all metrics, but in particular angular separation, to distinguish inter-personal from extra-personal face image differences. A novel technique, known as angularisation, is introduced by which a data set that is well separated in the angular sense can be mapped into a new feature space in which other metrics are equally discriminative. This operation can be performed separately or it can be incorporated into an SVM kernel. The benefit of angularisation is that it allows strong classification methods to take advantage of angular separation without explicitly incorporating it into their construction. It is shown that the accuracy of ECOC ensembles can be improved in this way. A further aspect of the research is to compare the effectiveness of the ECOC approach to constructing ensembles of SVM base classifiers with that of binary hierarchical classifiers (BHC). Experiments are performed which lead to the conclusion that, for face recognition problems, ECOC yields greater classification accuracy than the BHC method. This is attributed primarily to the fact that the size of the training set decreases along a path from the root node to a leaf node of the BHC tree and this leads to great difficulties in constructing accurate base classifiers at the lower nodes.
26
Lozano, Vega Gildardo. "Image-based detection and classification of allergenic pollen." Thesis, Dijon, 2015. http://www.theses.fr/2015DIJOS031/document.
Full textAbstract:
Le traitement médical des allergies nécessite la caractérisation des pollens en suspension dans l’air. Toutefois, cette tâche requiert des temps d’analyse très longs lorsqu’elle est réalisée de manière manuelle. Une approche automatique améliorerait ainsi considérablement les applications potentielles du comptage de pollens. Les dernières techniques d’analyse d’images permettent la détection de caractéristiques discriminantes. C’est pourquoi nous proposons dans cette thèse un ensemble de caractéristiques pertinentes issues d’images pour la reconnaissance des principales classes de pollen allergènes. Le cœur de notre étude est l’évaluation de groupes de caractéristiques capables de décrire correctement les pollens en termes de forme, texture, taille et ouverture. Les caractéristiques sont extraites d’images acquises classiquement sous microscope, permettant la reproductibilité de la méthode. Une étape de sélection des caractéristiques est appliquée à chaque groupe pour évaluer sa pertinence.Concernant les apertures présentes sur certains pollens, une méthode adaptative de détection, localisation et comptage pour différentes classes de pollens avec des apparences variées est proposée. La description des apertures se base sur une stratégie de type Sac-de-Mots appliquée à des primitives issues des images. Une carte de confiance est construite à partir de la confiance donnée à la classification des régions de l’image échantillonnée. De cette carte sont extraites des caractéristiques propres aux apertures, permettant leur comptage. La méthode est conçue pour être étendue de façon modulable à de nouveaux types d’apertures en utilisant le même algorithme mais avec un classifieur spécifique.Les groupes de caractéristiques ont été testés individuellement et conjointement sur les classes de pollens les plus répandues en Allemagne. Nous avons montré leur efficacité lors d’une classification de type SVM, notamment en surpassant la variance intra-classe et la similarité inter-classe. Les résultats obtenus en utilisant conjointement tous les groupes de caractéristiques ont abouti à une précision de 98,2 %, comparable à l’état de l’artThe correct classification of airborne pollen is relevant for medical treatment of allergies, and the regular manual process is costly and time consuming. An automatic processing would increase considerably the potential of pollen counting. Modern computer vision techniques enable the detection of discriminant pollen characteristics. In this thesis, a set of relevant image-based features for the recognition of top allergenic pollen taxa is proposed and analyzed. The foundation of our proposal is the evaluation of groups of features that can properly describe pollen in terms of shape, texture, size and apertures. The features are extracted on typical brightfield microscope images that enable the easy reproducibility of the method. A process of feature selection is applied to each group for the determination of relevance.Regarding apertures, a flexible method for detection, localization and counting of apertures of different pollen taxa with varying appearances is proposed. Aperture description is based on primitive images following the Bag-of-Words strategy. A confidence map is built from the classification confidence of sampled regions. From this map, aperture features are extracted, which include the count of apertures. The method is designed to be extended modularly to new aperture types employing the same algorithm to build individual classifiers.The feature groups are tested individually and jointly on of the most allergenic pollen taxa in Germany. They demonstrated to overcome the intra-class variance and inter-class similarity in a SVM classification scheme. The global joint test led to accuracy of 98.2%, comparable to the state-of-the-art procedures
27
Malkhare, Rohan V. "Scavenger: A Junk Mail Classification Program." Scholar Commons, 2003. https://scholarcommons.usf.edu/etd/1145.
Full textAbstract:
The problem of junk mail, also called spam, has reached epic proportions and various efforts are underway to fight spam. Junk mail classification using machine learning techniques is a key method to fight spam. We have devised a machine learning algorithm where features are created from individual sentences in the subject and body of a message by forming all possible word-pairings from a sentence. Weights are assigned to the features based on the strength of their predictive capabilities for spam/legitimate determination. The predictive capabilities are estimated by the frequency of occurrence of the feature in spam/legitimate collections as well as by application of heuristic rules. During classification, total spam and legitimate evidence in the message is obtained by summing up the weights of extracted features of each class and the message is classified into whichever class accumulates the greater sum.
We compared the algorithm against the popular naïve-bayes algorithm (in [8]) and found it's performance exceeded that of naïve-bayes algorithm both in terms of catching spam and for reducing false positives.
28
Saidi, Rabie. "Motif extraction from complex data : case of protein classification." Thesis, Clermont-Ferrand 2, 2012. http://www.theses.fr/2012CLF22272/document.
Full textAbstract:
La classification est l’un des défis important en bioinformatique, aussi bien pour les données protéiques que nucléiques. La présence de ces données en grandes masses, leur ambiguïté et en particulier les coûts élevés de l’analyse in vitro en termes de temps et d’argent, rend l’utilisation de la fouille de données plutôt une nécessité qu’un choix rationnel. Cependant, les techniques fouille de données, qui traitent souvent des données sous le format relationnel, sont confrontés avec le format inapproprié des données biologiques. Par conséquent, une étape inévitable de prétraitement doit être établie. Cette thèse traite du prétraitement de données protéiques comme une étape de préparation avant leur classification. Nous présentons l’extraction de motifs comme un moyen fiable pour répondre à cette tâche. Les motifs extraits sont utilisés comme descripteurs, en vue de coder les protéines en vecteurs d’attributs. Cela permet l’utilisation des classifieurs connus. Cependant, la conception d’un espace appropié d’attributs, n’est pas une tâche triviale. Nous traitons deux types de données protéiques à savoir les séquences et les structures 3D. Dans le premier axe, i:e:; celui des séquences, nous proposons un nouveau procédé de codage qui utilise les matrices de substitution d’acides aminés pour définir la similarité entre les motifs lors de l’étape d’extraction. En utilisant certains classifieurs, nous montrons l’efficacité de notre approche en la comparant avec plusieurs autres méthodes de codage. Nous proposons également de nouvelles métriques pour étudier la robustesse de certaines de ces méthodes lors de la perturbation des données d’entrée. Ces métriques permettent de mesurer la capacité d’une méthode de révéler tout changement survenant dans les données d’entrée et également sa capacité à cibler les motifs intéressants. Le second axe est consacré aux structures protéiques 3D, qui ont été récemment considérées comme graphes d’acides aminés selon différentes représentations. Nous faisons un bref survol sur les représentations les plus utilisées et nous proposons une méthode naïve pour aider à la construction de graphes d’acides aminés. Nous montrons que certaines méthodes répandues présentent des faiblesses remarquables et ne reflètent pas vraiment la conformation réelle des protéines. Par ailleurs, nous nous intéressons à la découverte, des sous-structures récurrentes qui pourraient donner des indications fonctionnelles et structurelles. Nous proposons un nouvel algorithme pour trouver des motifs spatiaux dans les protéines. Ces motifs obéissent à un format défini sur la base d’une argumentation biologique. Nous comparons avec des motifs séquentiels et spatiaux de certains travaux reliés. Pour toutes nos contributions, les résultats expérimentaux confirment l’efficacité de nos méthodes pour représenter les séquences et les structures protéiques, dans des tâches de classification. Les programmes développés sont disponibles sur ma page web http://fc.isima.fr/~saidiThe classification of biological data is one of the significant challenges inbioinformatics, as well for protein as for nucleic data. The presence of these data in hugemasses, their ambiguity and especially the high costs of the in vitro analysis in terms oftime and resources, make the use of data mining rather a necessity than a rational choice.However, the data mining techniques, which often process data under the relational format,are confronted with the inappropriate format of the biological data. Hence, an inevitablestep of pre-processing must be established.This thesis deals with the protein data preprocessing as a preparation step before theirclassification. We present motif extraction as a reliable way to address that task. The extractedmotifs are used as descriptors to encode proteins into feature vectors. This enablesthe use of known data mining classifiers which require this format. However, designing asuitable feature space, for a set of proteins, is not a trivial task.We deal with two kinds of protein data i:e:, sequences and tri-dimensional structures. In thefirst axis i:e:, protein sequences, we propose a novel encoding method that uses amino-acidsubstitution matrices to define similarity between motifs during the extraction step. Wedemonstrate the efficiency of such approach by comparing it with several encoding methods,using some classifiers. We also propose new metrics to study the robustness of some ofthese methods when perturbing the input data. These metrics allow to measure the abilityof the method to reveal any change occurring in the input data and also its ability to targetthe interesting motifs. The second axis is dedicated to 3D protein structures which are recentlyseen as graphs of amino acids. We make a brief survey on the most used graph-basedrepresentations and we propose a naïve method to help with the protein graph making. Weshow that some existing and widespread methods present remarkable weaknesses and do notreally reflect the real protein conformation. Besides, we are interested in discovering recurrentsub-structures in proteins which can give important functional and structural insights.We propose a novel algorithm to find spatial motifs from proteins. The extracted motifsmatch a well-defined shape which is proposed based on a biological basis. We compare withsequential motifs and spatial motifs of recent related works. For all our contributions, theoutcomes of the experiments confirm the efficiency of our proposed methods to representboth protein sequences and protein 3D structures in classification tasks.Software programs developed during this research work are available on my home page http://fc.isima.fr/~saidi
29
De, Voir Christopher S. "Wavelet Based Feature Extraction and Dimension Reduction for the Classification of Human Cardiac Electrogram Depolarization Waveforms." PDXScholar, 2005. https://pdxscholar.library.pdx.edu/open_access_etds/1740.
Full textAbstract:
An essential task for a pacemaker or implantable defibrillator is the accurate identification of rhythm categories so that the correct electrotherapy can be administered. Because some rhythms cause a rapid dangerous drop in cardiac output, it is necessary to categorize depolarization waveforms on a beat-to-beat basis to accomplish rhythm classification as rapidly as possible. In this thesis, a depolarization waveform classifier based on the Lifting Line Wavelet Transform is described. It overcomes problems in existing rate-based event classifiers; namely, (1) they are insensitive to the conduction path of the heart rhythm and (2) they are not robust to pseudo-events. The performance of the Lifting Line Wavelet Transform based classifier is illustrated with representative examples.
Although rate based methods of event categorization have served well in implanted devices, these methods suffer in sensitivity and specificity when atrial, and ventricular rates are similar. Human experts differentiate rhythms by morphological features of strip chart electrocardiograms. The wavelet transform is a simple approximation of this human expert analysis function because it correlates distinct morphological features at multiple scales. The accuracy of implanted rhythm determination can then be improved by using human-appreciable time domain features enhanced by time scale decomposition of depolarization waveforms.
The purpose of the present work was to determine the feasibility of implementing such a system on a limited-resolution platform. 78 patient recordings were split into equal segments of reference, confirmation, and evaluation sets. Each recording had a sampling rate of 512Hz, and a significant change in rhythm in the recording. The wavelet feature generator implemented in Matlab performs anti-alias pre-filtering, quantization, and threshold-based event detection, to produce indications of events to submit to wavelet transformation. The receiver operating characteristic curve was used to rank the discriminating power of the feature accomplishing dimension reduction. Accuracy was used to confirm the feature choice. Evaluation accuracy was greater than or equal to 95% over the IEGM recordings.
30
Stromann, Oliver. "Feature Extraction and FeatureSelection for Object-based LandCover Classification : Optimisation of Support Vector Machines in aCloud Computing Environment." Thesis, KTH, Geoinformatik, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-238727.
Full textAbstract:
Mapping the Earth’s surface and its rapid changes with remotely sensed data is a crucial tool to un-derstand the impact of an increasingly urban world population on the environment. However, the impressive amount of freely available Copernicus data is only marginally exploited in common clas-sifications. One of the reasons is that measuring the properties of training samples, the so-called ‘fea-tures’, is costly and tedious. Furthermore, handling large feature sets is not easy in most image clas-sification software. This often leads to the manual choice of few, allegedly promising features. In this Master’s thesis degree project, I use the computational power of Google Earth Engine and Google Cloud Platform to generate an oversized feature set in which I explore feature importance and analyse the influence of dimensionality reduction methods. I use Support Vector Machines (SVMs) for object-based classification of satellite images - a commonly used method. A large feature set is evaluated to find the most relevant features to discriminate the classes and thereby contribute most to high clas-sification accuracy. In doing so, one can bypass the sensitive knowledge-based but sometimes arbi-trary selection of input features.Two kinds of dimensionality reduction methods are investigated. The feature extraction methods, Linear Discriminant Analysis (LDA) and Independent Component Analysis (ICA), which transform the original feature space into a projected space of lower dimensionality. And the filter-based feature selection methods, chi-squared test, mutual information and Fisher-criterion, which rank and filter the features according to a chosen statistic. I compare these methods against the default SVM in terms of classification accuracy and computational performance. The classification accuracy is measured in overall accuracy, prediction stability, inter-rater agreement and the sensitivity to training set sizes. The computational performance is measured in the decrease in training and prediction times and the compression factor of the input data. I conclude on the best performing classifier with the most effec-tive feature set based on this analysis.In a case study of mapping urban land cover in Stockholm, Sweden, based on multitemporal stacks of Sentinel-1 and Sentinel-2 imagery, I demonstrate the integration of Google Earth Engine and Google Cloud Platform for an optimised supervised land cover classification. I use dimensionality reduction methods provided in the open source scikit-learn library and show how they can improve classification accuracy and reduce the data load. At the same time, this project gives an indication of how the exploitation of big earth observation data can be approached in a cloud computing environ-ment.The preliminary results highlighted the effectiveness and necessity of dimensionality reduction methods but also strengthened the need for inter-comparable object-based land cover classification benchmarks to fully assess the quality of the derived products. To facilitate this need and encourage further research, I plan to publish the datasets (i.e. imagery, training and test data) and provide access to the developed Google Earth Engine and Python scripts as Free and Open Source Software (FOSS).Kartläggning av jordens yta och dess snabba förändringar med fjärranalyserad data är ett viktigt verktyg för att förstå effekterna av en alltmer urban världsbefolkning har på miljön. Den imponerande mängden jordobservationsdata som är fritt och öppet tillgänglig idag utnyttjas dock endast marginellt i klassifikationer. Att hantera ett set av många variabler är inte lätt i standardprogram för bildklassificering. Detta leder ofta till manuellt val av få, antagligen lovande variabler. I det här arbetet använde jag Google Earth Engines och Google Cloud Platforms beräkningsstyrkan för att skapa ett överdimensionerat set av variabler i vilket jag undersöker variablernas betydelse och analyserar påverkan av dimensionsreducering. Jag använde stödvektormaskiner (SVM) för objektbaserad klassificering av segmenterade satellitbilder – en vanlig metod inom fjärranalys. Ett stort antal variabler utvärderas för att hitta de viktigaste och mest relevanta för att diskriminera klasserna och vilka därigenom mest bidrar till klassifikationens exakthet. Genom detta slipper man det känsliga kunskapsbaserade men ibland godtyckliga urvalet av variabler.Två typer av dimensionsreduceringsmetoder tillämpades. Å ena sidan är det extraktionsmetoder, Linjär diskriminantanalys (LDA) och oberoende komponentanalys (ICA), som omvandlar de ursprungliga variablers rum till ett projicerat rum med färre dimensioner. Å andra sidan är det filterbaserade selektionsmetoder, chi-två-test, ömsesidig information och Fisher-kriterium, som rangordnar och filtrerar variablerna enligt deras förmåga att diskriminera klasserna. Jag utvärderade dessa metoder mot standard SVM när det gäller exakthet och beräkningsmässiga prestanda.I en fallstudie av en marktäckeskarta över Stockholm, baserat på Sentinel-1 och Sentinel-2-bilder, demonstrerade jag integrationen av Google Earth Engine och Google Cloud Platform för en optimerad övervakad marktäckesklassifikation. Jag använde dimensionsreduceringsmetoder som tillhandahålls i open source scikit-learn-biblioteket och visade hur de kan förbättra klassificeringsexaktheten och minska databelastningen. Samtidigt gav detta projekt en indikation på hur utnyttjandet av stora jordobservationsdata kan nås i en molntjänstmiljö.Resultaten visar att dimensionsreducering är effektiv och nödvändig. Men resultaten stärker också behovet av ett jämförbart riktmärke för objektbaserad klassificering av marktäcket för att fullständigt och självständigt bedöma kvaliteten på de härledda produkterna. Som ett första steg för att möta detta behov och för att uppmuntra till ytterligare forskning publicerade jag dataseten och ger tillgång till källkoderna i Google Earth Engine och Python-skript som jag utvecklade i denna avhandling.
31
Koc, Bengi. "Detection And Classification Of Qrs Complexes From The Ecg Recordings." Master's thesis, METU, 2008. http://etd.lib.metu.edu.tr/upload/2/12610328/index.pdf.
Full textAbstract:
Electrocardiography (ECG) is the most important noninvasive tool used for diagnosing heart diseases. An ECG interpretation program can help the physician state the diagnosis correctly and take the corrective action. Detection of the QRS
complexes from the ECG signal is usually the first step for an interpretation tool. The main goal in this thesis was to develop robust and high performance QRS detection algorithms, and using the results of the QRS detection step, to classify these beats according to their different pathologies. In order to evaluate the performances, these algorithms were tested and compared in Massachusetts Institute of Technology Beth Israel Hospital (MIT-BIH) database, which was developed for research in cardiac electrophysiology.
In this thesis, four promising QRS detection methods were taken from literature and implemented: a derivative based method (Method I), a digital filter based method (Method II), Tompkin&rsquos method that utilizes the morphological features of the ECG signal (Method III) and a neural network based QRS detection method (Method IV). Overall sensitivity and positive predictivity values above 99% are achieved with each method, which are compatible with the results reported in literature. Method III has the best overall performance among the others with a sensitivity of 99.93% and a positive predictivity of 100.00%. Based on the detected QRS complexes, some features were extracted and classification of some beat types were performed. In order to classify the detected beats, three methods were taken from literature and implemented in this thesis: a Kth nearest neighbor rule based method (Method I), a neural network based method (Method II) and a rule based method (Method III). Overall results of Method I and Method II have sensitivity values above 92.96%. These findings are also compatible with those reported in the related literature. The classification made by the rule based approach, Method III, did not coincide well with the annotations provided in the MIT-BIH database. The best results were achieved by Method II with the overall sensitivity value of 95.24%.
32
Eklund, Martin. "Comparing Feature Extraction Methods and Effects of Pre-Processing Methods for Multi-Label Classification of Textual Data." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-231438.
Full textAbstract:
This thesis aims to investigate how different feature extraction methods applied to textual data affect the results of multi-label classification. Two different Bag of Words extraction methods are used, specifically the Count Vector and the TF-IDF approaches. A word embedding method is also investigated, called the GloVe extraction method. Multi-label classification can be useful for categorizing items, such as pieces of music or news articles, that may belong to multiple classes or topics. The effect of using different pre-processing methods is also investigated, such as the use of N-grams, stop-word elimination, and stemming. Two different classifiers, an SVM and an ANN, are used for multi-label classification using a Binary Relevance approach. The results indicate that the choice of extraction method has a meaningful impact on the resulting classifications, but that no one method consistently outperforms the others. Instead the results show that the GloVe extraction method performs the best for the recall metrics, while the Bag of Words methods perform the best for the precision metrics.Detta arbete ämnar att undersöka vilken effekt olika metoder för att extrahera särdrag ur textdata har när dessa används för att multi-tagga textdatan. Två metoder baserat på Bag of Words undersöks, närmare bestämt Count Vector-metoden samt TF-IDF-metoden. Även en metod som använder sig av word embessings undersöks, som kallas för GloVe-metoden. Multi-taggning av data kan vara användbart när datan, exempelvis musikaliska stycken eller nyhetsartiklar, kan tillhöra flera klasser eller områden. Även användandet av flera olika metoder för att förbehandla datan undersöks, såsom användandet utav N-gram, eliminering av icke-intressanta ord, samt transformering av ord med olika böjningsformer till gemensam stamform. Två olika klassificerare, en SVM samt en ANN, används för multi-taggningen genom använding utav en metod kallad Binary Relevance. Resultaten visar att valet av metod för extraktion av särdrag har en betydelsefull roll för den resulterande multi-taggningen, men att det inte finns en metod som ger bäst resultat genom alla tester. Istället indikerar resultaten att extraktionsmetoden baserad på GloVe presterar bäst när det gäller 'recall'-mätvärden, medan Bag of Words-metoderna presterar bäst gällade 'precision'-mätvärden.
33
Al-Qatawneh, Sokyna M. S. "3D Facial Feature Extraction and Recognition. An investigation of 3D face recognition: correction and normalisation of the facial data, extraction of facial features and classification using machine learning techniques." Thesis, University of Bradford, 2010. http://hdl.handle.net/10454/4876.
Full textAbstract:
Face recognition research using automatic or semi-automatic techniques has emerged over the last two decades. One reason for growing interest in this topic is the wide range of possible applications for face recognition systems. Another reason is the emergence of affordable hardware, supporting digital photography and video, which have made the acquisition of high-quality and high resolution 2D images much more ubiquitous. However, 2D recognition systems are sensitive to subject pose and illumination variations and 3D face recognition which is not directly affected by such environmental changes, could be used alone, or in combination with 2D recognition. Recently with the development of more affordable 3D acquisition systems and the availability of 3D face databases, 3D face recognition has been attracting interest to tackle the limitations in performance of most existing 2D systems. In this research, we introduce a robust automated 3D Face recognition system that implements 3D data of faces with different facial expressions, hair, shoulders, clothing, etc., extracts features for discrimination and uses machine learning techniques to make the final decision. A novel system for automatic processing for 3D facial data has been implemented using multi stage architecture; in a pre-processing and registration stage the data was standardized, spikes were removed, holes were filled and the face area was extracted. Then the nose region, which is relatively more rigid than other facial regions in an anatomical sense, was automatically located and analysed by computing the precise location of the symmetry plane. Then useful facial features and a set of effective 3D curves were extracted. Finally, the recognition and matching stage was implemented by using cascade correlation neural networks and support vector machine for classification, and the nearest neighbour algorithms for matching. It is worth noting that the FRGC data set is the most challenging data set available supporting research on 3D face recognition and machine learning techniques are widely recognised as appropriate and efficient classification methods.
34
Gao, Jiangning. "3D face recognition using multicomponent feature extraction from the nasal region and its environs." Thesis, University of Bath, 2016. https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.707585.
Full textAbstract:
This thesis is dedicated to extracting expression robust features for 3D face recognition. The use of 3D imaging enables the extraction of discriminative features that can significantly improve the recognition performance due to the availability of facial surface information such as depth, surface normals and curvature. Expression robust analysis using information from both depth and surface normals is investigated by dividing the main facial region into patches of different scales. The nasal region and adjoining parts of the cheeks are utilized as they are more consistent over different expressions and are hard to deliberately occlude. In addition, in comparison with other parts of the face, these regions have a high potential to produce discriminative features for recognition and overcome pose variations. An overview and classification methodology of the widely used 3D face databases are first introduced to provide an appropriate reference for 3D face database selection. Using the FRGC and Bosphorus databases, a low complexity pattern rejector for expression robust 3D face recognition is proposed by matching curves on the nasal and its environs, which results in a low-dimension feature set of only 60 points. To extract discriminative features more locally, a novel multi-scale and multi-component local shape descriptor is further proposed, which achieves more competitive performances under the identification and verification scenarios. In contrast with many of the existing work on 3D face recognition that consider captures obtained with laser scanners or structured light, this thesis also investigates applications to reconstructed 3D captures from lower cost photometric stereo imaging systems that have applications in real-world situations. To this end, the performance of the expression robust face recognition algorithms developed for captures from laser scanners are further evaluated on the Photoface database, which contains naturalistic expression variations. To improve the recognition performance of all types of 3D captures, a universal landmarking algorithm is proposed that makes uses of different components of the surface normals. Using facial profile signatures and thresholded surface normal maps, facial roll and yaw rotations are calibrated and five main landmarks are robustly detected on the well-aligned 3D nasal region. The landmarking results show that the detected landmarks demonstrate high within-class consistency and can achieve good recognition performances under different expressions. This is also the first landmarking work specifically developed for the reconstructed 3D captures from photometric stereo imaging systems.
35
Paradzinets, Aliaksandr V. "Variable resolution transform-based music feature extraction and their applications for music information retrieval." Ecully, Ecole centrale de Lyon, 2007. http://www.theses.fr/2007ECDL0047.
Full textAbstract:
Dans le secteur de loisirs il y a un nombre considérable d’enregistrements numériques musicaux produits, diffusés et échangés qui favorise la demande croisante de services intelligents de recherche de musique. La navigation par contenu devient cruciale pour permettre aux professionnels et également aux amateurs d’accéder facilement aux quantités de données musicales disponibles. Ce travail présente les nouveaux descripteurs de contenu musical et mesures de similarité qui permettent l’organisation automatique de données musicales (recherche par similarité, génération automatique des playlistes) ainsi que l’étiquetage (classification automatique en genres). Ce travail s’intéresse au problème de la construction des descripteurs du point de vue musical en complément des caractéristiques spectrales de bas-niveau. Plusieurs aspects d’analyse musicale, telles que l’analyse du signal où une nouvelle technique de transformation fréquentielle à résolution variable est proposée et décrite. Le traitement de niveau plus haut touche aux aspects de l’extraction des connaissances musicales. Cette thèse présente les algorithmes de détection de coups (beats) et d’extraction de fréquences fondamentales multiples. Les deux algorithmes sont basés sur la transformation à résolution variable proposée. Les informations issues de ces algorithmes sont utilisées dans la construction des descripteurs musicaux, représentés sous forme d’histogrammes (nouvel histogramme rythmique 2D qui permet d’estimer directement le tempo, et les histogrammes de succession et profil de notes). Deux applications majeures qui utilisent les caractéristiques mentionnées sont décrits et évaluées dans cette thèseAs a major product for entertainment, there is a huge amount of digital musical content produced, broadcasted, distributed and exchanged. There is a rising demand for content-based music search services. Similarity-based music navigation is becoming crucial for enabling easy access to the evergrowing amount of digital music available to professionals and amateurs alike. This work presents new musical content descriptors and similarity measures which allow automatic musical content organizing (search by similarity, automatic playlist generating) and labeling (automatic genre classification). The work considers the problem of content descriptor building from the musical point of view in complement of low-level spectral similarity measures. Several aspects of music analysis are considered such as music signal analysis where a novel variable resolution transform is presented and described. Higher level processing touches upon the musical knowledge extraction. The thesis presents algorithms of beat detection and multiple fundamental frequency estimation which are based on the variable resolution transform. The information issued from these algorithms is then used for building musical descriptors, represented in form of histograms (novel 2D beat histogram which enables a direct tempo estimation, note succession and note profile histograms etc. ). Two major music information retrieval applications, namely music genre classification and music retrieval by similarity, which use aforementioned musical features are described and evaluated in this thesis
36
Salmon, Brian Paxton. "Improved hyper-temporal feature extraction methods for land cover change detection in satellite time series." Thesis, University of Pretoria, 2012. http://hdl.handle.net/2263/28199.
Full textAbstract:
The growth in global population inevitably increases the consumption of natural resources. The need to provide basic services to these growing communities leads to an increase in anthropogenic changes to the natural environment. The resulting transformation of vegetation cover (e.g. deforestation, agricultural expansion, urbanisation) has significant impacts on hydrology, biodiversity, ecosystems and climate. Human settlement expansion is the most common driver of land cover change in South Africa, and is currently mapped on an irregular, ad hoc basis using visual interpretation of aerial photographs or satellite images. This thesis proposes several methods of detecting newly formed human settlements using hyper-temporal, multi-spectral, medium spatial resolution MODIS land surface reflectance satellite imagery. The hyper-temporal images are used to extract time series, which are analysed in an automated fashion using machine learning methods. A post-classification change detection framework was developed to analyse the time series using several feature extraction methods and classifiers. Two novel hyper-temporal feature extraction methods are proposed to characterise the seasonal pattern in the time series. The first feature extraction method extracts Seasonal Fourier features that exploits the difference in temporal spectra inherent to land cover classes. The second feature extraction method extracts state-space vectors derived using an extended Kalman filter. The extended Kalman filter is optimised using a novel criterion which exploits the information inherent in the spatio-temporal domain. The post-classification change detection framework was evaluated on different classifiers; both supervised and unsupervised methods were explored. A change detection accuracy of above 85% with false alarm rate below 10% was attained. The best performing methods were then applied at a provincial scale in the Gauteng and Limpopo provinces to produce regional change maps, indicating settlement expansion.Thesis (PhD(Eng))--University of Pretoria, 2012.
Electrical, Electronic and Computer Engineering
unrestricted
37
Gashayija, Jean Marie. "Image classification, storage and retrieval system for a 3 u cubesat." Thesis, Cape Peninsula University of Technology, 2014. http://hdl.handle.net/20.500.11838/1189.
Full textAbstract:
Thesis submitted in fulfillment of the requirements for the degree Master of Technology: Electrical Engineering in the Faculty of Engineering at the Cape Peninsula University of TechnologySmall satellites, such as CubeSats are mainly utilized for space and earth imaging missions. Imaging CubeSats are equipped with high resolution cameras for the capturing of digital images, as well as mass storage devices for storing the images. The captured images are transmitted to the ground station and subsequently stored in a database. The main problem with stored images in a large image database, identified by researchers and developers within the last number of years, is the retrieval of precise, clear images and overcoming the semantic gap. The semantic gap relates to the lack of correlation between the semantic categories the user requires and the low level features that a content-based image retrieval system offers. Clear images are needed to be usable for applications such as mapping, disaster monitoring and town planning. The main objective of this thesis is the design and development of an image classification, storage and retrieval system for a CubeSat. This system enables efficient classification, storing and retrieval of images that are received on a daily basis from an in-orbit CubeSat. In order to propose such a system, a specific research methodology was chosen and adopted. This entails extensive literature reviews on image classification techniques and image feature extraction techniques, to extract content embedded within an image, and include studies on image database systems, data mining techniques and image retrieval techniques. The literature study led to a requirement analysis followed by the analyses of software development models in order to design the system. The proposed design entails classifying images using content embedded in the image and also extracting image metadata such as date and time. Specific features extraction techniques are needed to extract required content and metadata. In order to achieve extraction of information embedded in the image, colour feature (colour histogram), shape feature (Mathematical Morphology) and texture feature (GLCM) techniques were used. Other major contributions of this project include a graphical user interface which enables users to search for similar images against those stored in the database. An automatic image extractor algorithm was also designed to classify images according to date and time, and colour, texture and shape features extractor techniques were proposed. These ensured that when a user wishes to query the database, the shape objects, colour quantities and contrast contained in an image are extracted and compared to those stored in the database. Implementation and test results concluded that the designed system is able to categorize images automatically and at the same time provide efficient and accurate results. The features extracted for each image depend on colour, shape and texture methods. Optimal values were also incorporated in order to reduce retrieval times. The mathematical morphological technique was used to compute shape objects using erosion and dilation operators, and the co-occurrence matrix was used to compute the texture feature of the image.
38
Avan, Selcuk Kazim. "Feature Set Evaluation For A Generic Missile Detection System." Master's thesis, METU, 2007. http://etd.lib.metu.edu.tr/upload/2/12608130/index.pdf.
Full textAbstract:
Missile Detection System (MDS) is one of the main components of a self-protection system developed against the threat of guided missiles for airborne platforms. The requirements such as time critical operation and high accuracy in classification performance make the &lsquoPattern Recognition&rsquo
problem of an MDS a hard task. Problem can be defined in two main parts such as &lsquo
Feature Set Evaluation&rsquo
(FSE) and &lsquo
Classifier&rsquo
designs. The main goal of feature set evaluation is to employ a dimensionality reduction process for the input data set, while not disturbing the classification performance in the result. In this thesis study, FSE approaches are investigated for the pattern recognition problem of a generic MDS. First, synthetic data generation is carried out in software environment by employing generic models and assumptions in order to reflect the nature of a realistic problem environment. Then, data sets are evaluated in order to draw a baseline for further feature set evaluation approaches. Further, a theoretical background including the concepts of Class Separability, Feature Selection and Feature Extraction is given. Several widely used methods are assessed in terms of convenience for the problem by giving necessary justifications depending on the data set characteristics. Upon this background, software implementations are performed regarding several feature set evaluation techniques. Simulations are carried out in order to process dimensionality reduction. For the evaluation of the resulting data sets in terms of classification performance, software implementation of a classifier is realized. Resulting classification performances of the applied approaches are compared and evaluated.
39
Wang, Xuechuan, and n/a. "Feature Extraction and Dimensionality Reduction in Pattern Recognition and Their Application in Speech Recognition." Griffith University. School of Microelectronic Engineering, 2003. http://www4.gu.edu.au:8080/adt-root/public/adt-QGU20030619.162803.
Full textAbstract:
Conventional pattern recognition systems have two components: feature analysis and pattern classification. Feature analysis is achieved in two steps: parameter extraction step and feature extraction step. In the parameter extraction step, information relevant for pattern classification is extracted from the input data in the form of parameter vector. In the feature extraction step, the parameter vector is transformed to a feature vector. Feature extraction can be conducted independently or jointly with either parameter extraction or classification. Linear Discriminant Analysis (LDA) and Principal Component Analysis (PCA) are the two popular independent feature extraction algorithms. Both of them extract features by projecting the parameter vectors into a new feature space through a linear transformation matrix. But they optimize the transformation matrix with different intentions. PCA optimizes the transformation matrix by finding the largest variations in the original feature space. LDA pursues the largest ratio of between-class variation and within-class variation when projecting the original feature space to a subspace. The drawback of independent feature extraction algorithms is that their optimization criteria are different from the classifiers minimum classification error criterion, which may cause inconsistency between feature extraction and the classification stages of a pattern recognizer and consequently, degrade the performance of classifiers. A direct way to overcome this problem is to conduct feature extraction and classification jointly with a consistent criterion. Minimum classification Error (MCE) training algorithm provides such an integrated framework. MCE algorithm was first proposed for optimizing classifiers. It is a type of discriminative learning algorithm but achieves minimum classification error directly. The flexibility of the framework of MCE algorithm makes it convenient to conduct feature extraction and classification jointly. Conventional feature extraction and pattern classification algorithms, LDA, PCA, MCE training algorithm, minimum distance classifier, likelihood classifier and Bayesian classifier, are linear algorithms. The advantage of linear algorithms is their simplicity and ability to reduce feature dimensionalities. However, they have the limitation that the decision boundaries generated are linear and have little computational flexibility. SVM is a recently developed integrated pattern classification algorithm with non-linear formulation. It is based on the idea that the classification that a.ords dot-products can be computed efficiently in higher dimensional feature spaces. The classes which are not linearly separable in the original parametric space can be linearly separated in the higher dimensional feature space. Because of this, SVM has the advantage that it can handle the classes with complex nonlinear decision boundaries. However, SVM is a highly integrated and closed pattern classification system. It is very difficult to adopt feature extraction into SVMs framework. Thus SVM is unable to conduct feature extraction tasks. This thesis investigates LDA and PCA for feature extraction and dimensionality reduction and proposes the application of MCE training algorithms for joint feature extraction and classification tasks. A generalized MCE (GMCE) training algorithm is proposed to mend the shortcomings of the MCE training algorithms in joint feature and classification tasks. SVM, as a non-linear pattern classification system is also investigated in this thesis. A reduced-dimensional SVM (RDSVM) is proposed to enable SVM to conduct feature extraction and classification jointly. All of the investigated and proposed algorithms are tested and compared firstly on a number of small databases, such as Deterding Vowels Database, Fishers IRIS database and Germans GLASS database. Then they are tested in a large-scale speech recognition experiment based on TIMIT database.
40
Doo, Seung Ho. "Analysis, Modeling & Exploitation of Variability in Radar Images." The Ohio State University, 2016. http://rave.ohiolink.edu/etdc/view?acc_num=osu1461256996.
Full text
41
Khodjet-Kesba, Mahmoud. "Automatic target classification based on radar backscattered ultra wide band signals." Thesis, Clermont-Ferrand 2, 2014. http://www.theses.fr/2014CLF22506/document.
Full textAbstract:
L’objectif de cette thèse est la classification automatique des cibles (ATC) en utilisant les signaux rétrodiffusés par un radar ultra large bande (UWB). La classification des cibles est réalisée en comparant les signatures des cibles et les signatures stockées dans une base de données. Premièrement, une étude sur la théorie de diffusion nous a permis de comprendre le sens physique des paramètres extraits et de les exprimer mathématiquement. Deuxièmement, des méthodes d’extraction de paramètres sont appliquées afin de déterminer les signatures des cibles. Un bon choix des paramètres est important afin de distinguer les différentes cibles. Différentes méthodes d’extraction de paramètres sont comparées notamment : méthode de Prony, Racine-classification des signaux multiples (Root-MUSIC), l’estimation des paramètres des signaux par des techniques d’invariances rotationnels (ESPRIT), et la méthode Matrix Pencil (MPM). Troisièmement, une méthode efficace de classification supervisée est nécessaire afin de classer les cibles inconnues par l’utilisation de leurs signatures extraites. Différentes méthodes de classification sont comparées notamment : Classification par la distance de Mahalanobis (MDC), Naïve Bayes (NB), k-plus proches voisins (k-NN), Machines à Vecteurs de Support (SVM). Une bonne technique de classification doit avoir une bonne précision en présence de signaux bruités et quelques soit l’angle d’émission. Les différents algorithmes ont été validés en utilisant les simulations des données rétrodiffusées par des objets canoniques et des cibles de géométries complexes modélisées par des fils minces et parfaitement conducteurs. Une méthode de classification automatique de cibles basée sur l’utilisation de la méthode Matrix Pencil dans le domaine fréquentiel (MPMFD) pour l’extraction des paramètres et la classification par la distance de Mahalanobis est proposée. Les résultats de simulation montrent que les paramètres extraits par MPMFD présentent une solution plausible pour la classification automatique des cibles. En outre, nous avons prouvé que la méthode proposée a une bonne tolérance aux bruits lors de la classification des cibles. Enfin, les différents algorithmes sont validés sur des données expérimentales et cibles réellesThe objective of this thesis is the Automatic Target Classification (ATC) based on radar backscattered Ultra WideBand (UWB) signals. The classification of the targets is realized by making comparison between the deduced target properties and the different target features which are already recorded in a database. First, the study of scattering theory allows us to understand the physical meaning of the extracted features and describe them mathematically. Second, feature extraction methods are applied in order to extract signatures of the targets. A good choice of features is important to distinguish different targets. Different methods of feature extraction are compared including wavelet transform and high resolution techniques such as: Prony’s method, Root-Multiple SIgnal Classification (Root-MUSIC), Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) and Matrix Pencil Method (MPM). Third, an efficient method of supervised classification is necessary to classify unknown targets by using the extracted features. Different methods of classification are compared: Mahalanobis Distance Classifier (MDC), Naïve Bayes (NB), k-Nearest Neighbors (k-NN) and Support Vector Machine (SVM). A useful classifier design technique should have a high rate of accuracy in the presence of noisy data coming from different aspect angles. The different algorithms are demonstrated using simulated backscattered data from canonical objects and complex target geometries modeled by perfectly conducting thin wires. A method of ATC based on the use of Matrix Pencil Method in Frequency Domain (MPMFD) for feature extraction and MDC for classification is proposed. Simulation results illustrate that features extracted with MPMFD present a plausible solution to automatic target classification. In addition, we prove that the proposed method has better ability to tolerate noise effects in radar target classification. Finally, the different algorithms are validated on experimental data and real targets
42
Ersoy, Mehmet Okan. "Application Of A Natural-resonance Based Feature Extraction Technique To Small-scale Aircraft Modeled By Conducting Wires For Electromagnetic Target Classification." Master's thesis, METU, 2004. http://etd.lib.metu.edu.tr/upload/3/12605522/index.pdf.
Full textAbstract:
The problem studied in this thesis, is the classification of the small-scale
aircraft targets by using a natural resonance based electromagnetic feature extraction
technique. The aircraft targets are modeled by perfectly conducting, thin wire
structures. The electromagnetic back-scattered data used in the classification process,
are numerically generated for five aircraft models.
A contemporary signal processing tool, the Wigner-Ville distribution is
employed in this study in addition to using the principal components analysis
technique to extract target features mainly from late-time target responses. The
Wigner-Ville distribution (WD) is applied to the electromagnetic back-scattered
responses from different aspects. Then, feature vectors are extracted from suitably
chosen late-time portions of the WD outputs, which include natural resonance related
v
information, for every target and aspect to decrease aspect dependency. The database
of the classifier is constructed by the feature vectors extracted at only a few reference
aspects. Principal components analysis is also used to fuse the feature vectors and/or
late-time aircraft responses extracted from reference aspects of a given target into a
single characteristic feature vector of that target to further reduce aspect dependency.
Consequently, an almost aspect independent classifier is designed for small-scale
aircraft targets reaching high correct classification rate.
43
Kachouri, Rostom. "Classification multi-modèles des images dans les bases Hétérogènes." Phd thesis, Université d'Evry-Val d'Essonne, 2010. http://tel.archives-ouvertes.fr/tel-00526649.
Full textAbstract:
La reconnaissance d'images est un domaine de recherche qui a été largement étudié par la communauté scientifique. Les travaux proposés dans ce cadre s'adressent principalement aux diverses applications des systèmes de vision par ordinateur et à la catégorisation des images issues de plusieurs sources. Dans cette thèse, on s'intéresse particulièrement aux systèmes de reconnaissance d'images par le contenu dans les bases hétérogènes. Les images dans ce type de bases appartiennent à différents concepts et représentent un contenu hétérogène. Pour ce faire, une large description permettant d'assurer une représentation fiable est souvent requise. Cependant, les caractéristiques extraites ne sont pas nécessairement toutes appropriées pour la discrimination des différentes classes d'images qui existent dans une base donnée d'images. D'où, la nécessité de sélection des caractéristiques pertinentes selon le contenu de chaque base. Dans ce travail, une méthode originale de sélection adaptative est proposée. Cette méthode permet de considérer uniquement les caractéristiques qui sont jugées comme les mieux adaptées au contenu de la base d'image utilisée. Par ailleurs, les caractéristiques sélectionnées ne disposent pas généralement des mêmes performances. En conséquence, l'utilisation d'un algorithme de classification, qui s'adapte aux pouvoirs discriminants des différentes caractéristiques sélectionnées par rapport au contenu de la base d'images utilisée, est vivement recommandée. Dans ce contexte, l'approche d'apprentissage par noyaux multiples est étudiée et une amélioration des méthodes de pondération des noyaux est présentée. Cette approche s'avère incapable de décrire les relations non-linéaires des différents types de description. Ainsi, nous proposons une nouvelle méthode de classification hiérarchique multi-modèles permettant d'assurer une combinaison plus flexible des caractéristiques multiples. D'après les expérimentations réalisées, cette nouvelle méthode de classification assure des taux de reconnaissance très intéressants. Enfin, les performances de la méthode proposée sont mises en évidence à travers une comparaison avec un ensemble d'approches cité dans la littérature récente du domaine.
44
Chiou, Tzone-Kaie, and 邱宗楷. "Using Fuzzy Feature Extraction Fingerprint Classification." Thesis, 2001. http://ndltd.ncl.edu.tw/handle/92760623300716087147.
Full textAbstract:
碩士元智大學
資訊工程學系
89
Fingerprint classification is a useful task for a large database of fingerprint recognition system. Accurate classification can speed up the process of fingerprint recognition. The fingerprint classification method proposed in this paper is based on human thinking and uses fuzzy theory. The key point of human thinking to classify fingerprint is attempting to find out fingerprint ridge, singular points (cores or deltas), direction of ridge, wrinkles or scars as global features. Firstly, in order to determine the fingerprint ridge direction, we need to transform the fingerprint image into 50x50 direction pattern. Then we use a set of pre-defined fuzzy mask to find out the singular points. Finally we use relationship between the singular points to classify the fingerprint. The experimental results of our method exhibit the best performance, a very low sensitivity and good classification accuracy.
45
Jiun-Jin, Huang, and 黃俊錦. "Effects of Feature Extraction on Classification Accuracy." Thesis, 1998. http://ndltd.ncl.edu.tw/handle/86200872473956722918.
Full textAbstract:
碩士國立臺灣科技大學
管理技術研究所
86
Classification is an important area in pattern recognition. Feature extra ction for classification is equivalent to retaining informative features or eliminating redundant features. However, due to the nonlinearity of the decision boundary, which occurs in most cases, there exist no absolutely but approxima tely redundant features. Eliminating approximately redundant features results in a decrease in the classification accuracy. Even for two classes with multiv ariate normal distributions, classification accuracy is difficult to analyze s ince the classification function involves quadratic terms. One approach to all eviating this difficulty is to simultaneously diagonalize the covariance matri ces of the two classes which can be achieved by applying orthornormal and whit ening transformations to the measurement space. Once the covariance matrices are simultaneously diagonalized, the quadratic classification function is simplified and becomes much easier to analyze and the classification accuracy can be studied in terms of the eigenvalues of the covariance matrices of the two classes. Thus, the decrease in the classification accuracy incurred from eliminating approximately redundant features can be quantified.We empirically study the classification accuracy by varying the distributionparameters.
46
Najdi, Shirin. "Feature Extraction and Selection in Automatic Sleep Stage Classification." Doctoral thesis, 2018. http://hdl.handle.net/10362/66271.
Full textAbstract:
Sleep stage classification is vital for diagnosing many sleep related
disorders and Polysomnography (PSG) is an important tool in this regard.
The visual process of sleep stage classification is time consuming, subjective
and costly. To improve the accuracy and efficiency of the sleep stage
classification, researchers have been trying to develop automatic
classification algorithms.
The automatic sleep stage classification mainly consists of three steps:
pre-processing, feature extraction and classification. In this research work,
we focused on feature extraction and selection steps. The main goal of this
thesis was identifying a robust and reliable feature set that can lead to
efficient classification of sleep stages. For achieving this goal, three types of
contributions were introduced in feature selection, feature extraction and
feature vector quality enhancement.
Several feature ranking and rank aggregation methods were evaluated and
compared for finding the best feature set. Evaluation results indicated that
the decision on the precise feature selection method depends on the system
design requirements such as low computational complexity, high stability
or high classification accuracy. In addition to conventional feature ranking
methods, in this thesis, novel methods such as Stacked Sparse AutoEncoder
(SSAE) was used for dimensionality reduction.
In feature extration area, new and effective features such as distancebased
features were utilized for the first time in sleep stage classification.
The results showed that these features contribute positively to the
classification performance. For signal quality enhancement, a loss-less EEG
artefact removal algorithm was proposed. The proposed adaptive algorithm
led to a significant enhancement in the overall classification accuracy.
47
Lin, Chia-Hsing, and 林家興. "Discriminative Feature Extraction for Robust Audio Event Classification." Thesis, 2010. http://ndltd.ncl.edu.tw/handle/4v582f.
Full textAbstract:
碩士國立臺北科技大學
電腦與通訊研究所
98
In Tradition, audio event classification relies heavily on MFCCs (Mel-Frequency Cepstral Coefficients) features. However, MFCCs is originally designed for automatic speech recognition. It is not sure whether MFCCs are still the best features for audio event classification or not. Besides, MFCCs are usually not so robust in noisy environment. Therefore, in this paper, several new feature extraction methods are proposed in the hope of getting better performance and robustness than MFCCs in noisy conditions. The proposed feature extraction methods are mainly based on the concept of match filters in spectro-temporal domain. Several methods to design the set of match filters are proposed including handmade gabor filters and three data-driven filters using PCA (Principle Component Analysis), LDA-based Eigen-space analysis (Linear Discriminative Analysis) and MCE (Minimum Classification Error) training. The robustness of the proposed method is evaluated on RWCP (Real World Computing Partnership) database with artificially added noise. There are 105 different audio events in RWCP. The experimental settings are similar to Aurora 2 multi-condition training task. Experimental results show that the lowest average error rate of 3.17% was achieved by MCE method and is superior to conventional MFCCs (4.13%). We thus confirm the superiority and robustness of the proposed audio feature extraction approaches.
48
Liu, Yu-Hsin, and 劉羽欣. "Feature extraction and classification of product advertising review." Thesis, 2013. http://ndltd.ncl.edu.tw/handle/q67wnd.
Full textAbstract:
碩士國立臺北科技大學
資訊工程系研究所
101
Web has become an important place for marketing in business. Many vendors offer bloggers or people their products or payment and ask them to write review of product using experience to promote their products. However, it’s hard to identify the truthfulness of these reviews. By using conventional text classification methods by content, it is difficult to distinguish between real and fake reviews. In this paper, we propose a feature extraction method and classification model for advertising reviews. Based on features like ratio of positive opinion terms, number of pictures, ratio of praiseful words, and publishes date; we train a SVM classifier for advertising review identification. In our experiment, we collected 2150 reviews in the “cosmetics” domain. For classifying advertising reviews in cosmetics domain and other articles, our method can perform 94% at F-measure. This result is comparable to the conventional approach of document classification using TF-IDF, and our method is more efficient in training. For classifying advertising and ordinary non-advertising reviews in cosmetics domain, our method also can achieve good classification accuracy. It shows the feasibility of practical use in advertising reviews classification.
49
Huang, Bin. "Compression ECG signals with feature extraction, classification, and browsability." 2004. http://hdl.handle.net/1993/16253.
Full text
50
Li, Ting-Yi, and 李庭誼. "Hybrid Feature Extraction for Object-based Hyperspectral Image Classification." Thesis, 2011. http://ndltd.ncl.edu.tw/handle/46541455753533757818.
Full textAbstract:
碩士國立臺灣大學
土木工程學研究所
99
The purpose of feature extraction is to reduce the dimensionality of hyperspectral images to solve classification problems caused by limited training samples. In this study, a hybrid feature extraction method which integrates spectral features and spatial features simultaneously is proposed. Firstly, the spectral-feature images are calculated along the spectral dimension of hyperspectral images using wavelet decomposition because wavelet has been proven effective in extracting spectral features. Secondly, ten different kinds of spatial-features, which are calculated along the two spatial dimensions of hyperspectral images, are implemented on the wavelet spectral-feature images. Then a feature selection method based on the optimization of class separability is performed on the extracted spectral-spatial features to get the hybrid features which could be suitable for classification applications. In this study, the object-based image analysis (OBIA) is used for hyperspectral image classification. The experiment results showed that the overall accuracy for the classification of a real hyperspectral data set using our proposed approach could reach approximately 94%. Moreover, it is worth mentioning that the hybrid features and OBIA classification could significantly rise the overall accuracy of hyperspectral images which contain poor separability between classes, after the spectral features were extracted. The experiment result also showed that the overall accuracy would go up by 20% by using our proposed approach on hyperspectral images with poor class separability.