Accurate methods for glottal feature extraction include the use of high-speed video imaging (HSVI). There have been previous attempts to extract these features with the acoustic recording. However, none of these methods compare their results with an objective method, such as HSVI. This thesis tests these acoustic methods against a large diverse population of 46 subjects. Two previously studied acoustic methods, as well as one introduced in this thesis, were compared against two video methods, area and displacement for open quotient (OQ) estimation. The area comparison proved to be somewhat ambiguous and challenging due to thresholding effects. The displacement comparison, which is based on glottal edge tracking, proved to be a more robust comparison method than the area. The first acoustic methods OQ estimate had a relatively small average error of 8.90% and the second method had a relatively large average error of -59.05% compared to the displacement OQ. The newly proposed method had a relatively small error of -13.75% when compared to the displacements OQ. There was some success even though there was relatively high error with the acoustic methods, however, they may be utilized to augment the features collected by HSVI for a more accurate glottal feature estimation.

In this thesis we explore robust statistical syntax analysis for French. Our main concern is to explore methods whereby the linguist can inject linguistic knowledge and/or resources into the robust statistical engine in order to improve results for specific phenomena. We first explore the dependency annotation schema for French, concentrating on certain phenomena. Next, we look into the various algorithms capable of producing this annotation, and in particular on the transition-based parsing algorithm used in the rest of this thesis. After exploring supervised machine learning algorithms for NLP classification problems, we present the Talismane toolkit for syntax analysis, built within the framework of this thesis, including four statistical modules - sentence boundary detection, tokenisation, pos-tagging and parsing - as well as the various linguistic resources used for the baseline model, including corpora, lexicons and feature sets. Our first experiments attempt various machine learning configurations in order to identify the best baseline. We then look into improvements made possible by beam search and beam propagation. Finally, we present a series of experiments aimed at correcting errors related to specific linguistic phenomena, using targeted features. One our innovation is the introduction of rules that can impose or prohibit certain decisions locally, thus bypassing the statistical model. We explore the usage of rules for errors that the features are unable to correct. Finally, we look into the enhancement of targeted features by large scale linguistic resources, and in particular a semi-supervised approach using a distributional semantic resource.

Bakalauro darbe analizuojami specialiojo pedagogo veiklos ypatumai mokyklos vaiko gerovės komisijoje. Tyrime dalyvavo bendrojo lavinimo mokyklos vaiko gerovės komisijos nariai iš Vilniaus rajono (6 specialieji pedagogai, 2 logopedai, 1 psichologas, 1 socialinis pedagogas, 2 pavaduotojos ugdymui). Teorinė darbo dalis atskleidžia specialiosios pedagoginės pagalbos ypatumus šiuolaikinėje mokykloje, parodo vaiko gerovės komisijos vaidmenį koordinuojant specialiąją pedagoginę pagalbą ir užtikrinant jos veiksmingumą, nustato specialiojo pedagogo funkcijos mokykloje, pristato specialųjį pedagogą kaip vaiko gerovės komisijos komandos narį. Naudojant interviu metodą, atskleidžiami specialiojo pedagogo veiklos parametrai, bei jo veiklos struktūros komponentai vaiko gerovės komisijoje. Taip pat vaiko gerovės komisijos narių požiūris į specialiosios pedagoginės pagalbos veiksmingumą mokyklos vaiko gerovės komisijoje. Analizuojant Vilniaus rajono bendrojo ugdymo mokyklų atsakymų kokybinio tyrimo rezultatus, buvo pastebėta, kad vaiko gerovės komisijos nariai specialiojo pedagogo veiklos parametrus vaiko gerovės komisijoje supranta ir apibūdina kaip platų, daugialypį procesą. Apklausti informantai teigia, jog specialiųjų pedagogų veiklos sritis ir pobūdis yra labai platus. Išryškėja tokios veiklos sritys, kaip darbas su SUP mokiniais, mokytojais, tėvais/šeima/įtėviais, darbas komandoje, tai pat išryškėja dar viena veiklos sritis – dokumentų tvarkymas, o kalbant apie specialiojo pedagogo... [toliau žr. visą tekstą] / This thesis analyzes the distinctive features of the Special Education Teacher's activities in the School’s Children Welfare Council. The research was made using the interview method. Interviews were conducted with members of Children Welfare Commission from different basic education schools of the Vilnius district (6 special education teachers, 2 speech therapists, 1 psychologist, 1 social educator, 2 deputy directors for educational matters). The theoretical part reveals the features of special educational support within a contemporary school, demonstrates the role of the School’s Children Welfare Commission in coordinating special educational support measures as well as in ensuring their efficiency. It also discusses the functions of the Special Education Teacher within a school, introduces her (him) as a member of the School’s Children Welfare Council. The interviews conducted reveal the parameters of a Special Education Teacher as well as the components being part of the structure of the activity within the School’s Children Welfare Council. They also provide an evaluation of the efficiency of special educational support within the School’s Children Welfare Council by the council members. The results of the qualitative research of the answers made by representatives of schools of the Vilnius district showed that the council members perceive and describe the parameters of the Special Education Teachers within the School’s Children Welfare Council as a broad and... [to full text]

Research based on translation expertise, which is also sometimes referred to as translation competence, has been a growing area of investigation in translation studies. These studies have not only focused on how translation expertise may be conceptualised and defined, but also on how this expertise is acquired and developed by translators. One of the key observations that arise from an overview of current research in the field of translation expertise is the prevalence of process-oriented methodologies in the field, with product-oriented methodologies used comparatively infrequently. This study is based on the assumption that product-oriented methodologies, and specifically the corpus-based approach, may provide new insights into translation expertise. The study therefore sets out to address the lack of comprehensive and systematic corpus-based analyses of translation expertise. One of the foremost concerns of corpus-based translation studies has been the investigation of what is known as the features of translated language which are often categorised as: explicitation, simplification, normalisation and levelling-out. The main objective of this study is to investigate the hypothesis that the features of translated language can be taken as an index of translation expertise. The hypothesis is founded on the premise that if the features of translated language are considered to be the textual traces of translation strategies, then the different translation strategies associated with different levels of translation expertise will be reflected in different frequencies and distributions of these features of translated language in the work of experienced and inexperienced translators. The study therefore aimed to determine if there are significant differences in the frequency and distribution of the features of translated language in the work of experienced and inexperienced translators. As background to this main research question, the study also investigated a secondary hypothesis in which translated language demonstrates unique features that are the consequence of various aspects of the translation process. A custom-built comparable English corpus was used for the study, comprising three subcorpora: translations by experienced translators, translations by inexperienced translators, and non-translations. A selection of linguistic operationalization’s was chosen for each of the four features of translated language. The differences in the frequency and distribution of these linguistic operationalization’s in the three sub corpora were analysed by means of parametric or non-parametric ANOVA. The findings of the study provide some support for both hypotheses. In terms of the translation expertise hypothesis, some of the features of translated language demonstrate significantly different frequencies in the work of experienced translators compared to the work of inexperienced translators. It was found that experienced translators are less explicit in terms of: formal completeness, simplify less frequently because they use a more varied vocabulary, use longer sentences and have a lower readability index score on their translations, and use contractions more frequently, which signals that they normalise less than inexperienced translators. However, experienced translators also use neologisms and loanwords less frequently than inexperienced translators, which is suggestive of normalisation occurring more often in the work of experienced translators when it comes to lexical creativity. These linguistic differences are taken as indicative of the different translation strategies used by the two groups of translators. It is believed that the differences are primarily caused by variations in experienced and inexperienced translators‟ sensitivity to translation norms, their awareness of written language conventions, their language competence (which involves syntactic, morphological and vocabulary knowledge), and their sensitivity to register. Furthermore, it was also found that there are indeed significant differences between translated and non-translated language, which also provides support for the second hypothesis investigated in this study. Translators explicitate more frequently than non-translators in terms of formal completeness, tend to have a less extensive vocabulary, tend to raise the overall formality of their translations, and produce texts that are less creative and more conformist than non-translators‟ texts. However, statistical support is lacking for the hypothesis that translators explicitate more at the propositional level than original text producers do, as well as for the hypothesis that translators are inclined to use a more neutral middle register. / MA (Language Practice), North-West University, Vaal Triangle Campus, 2013

We present an algorithm for imputation of incomplete datasets based on Bayesian exchangeability through Pólya sampling. Each (donee) unit with a missing value is imputed multiple times by observed (real) values on units from a donor pool. The donor pools are constructed using auxiliary variables. Several features from kernel estimation are used to counteract unbalances that are due to sparse and bounded data. Three balancing features can be used with only one single continuous auxiliary variable, but an additional fourth feature need, multiple continuous auxiliary variables. They mainly contribute by reducing nonresponse bias. We examine how the donor pool size should be determined, that is the number of potential donors within the pool. External information is shown to be easily incorporated in the imputation algorithm. Our simulation studies show that with a study variable which can be seen as a function of one or two continuous auxiliaries plus residual noise, the method performs as well or almost as well as competing methods when the function is linear, but usually much better when the function is nonlinear. / <p>At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 1: In press. Paper 3: Submitted. Paper 4: Submitted.</p>

This thesis presents a framework for automatic recognition of human actions in uncontrolled, realistic video data such as movies, internet and surveillance videos. In this thesis, the human action recognition problem is solved from the perspective of local spatio-temporal feature and bag-of-features representation. The bag-of-features model only contains statistics of unordered low-level primitives, and any information concerning temporal ordering and spatial structure is lost. To address this issue, we proposed a novel multiscale local part model on the purpose of maintaining both structure information and ordering of local events for action recognition. The method includes both a coarse primitive level root feature covering event-content statistics and higher resolution overlapping part features incorporating local structure and temporal relationships. To extract the local spatio-temporal features, we investigated a random sampling strategy for efficient action recognition. We also introduced the idea of using very high sampling density for efficient and accurate classification. We further explored the potential of the method with the joint optimization of two constraints: the classification accuracy and its efficiency. On the performance side, we proposed a new local descriptor, called GBH, based on spatial and temporal gradients. It significantly improved the performance of the pure spatial gradient-based HOG descriptor on action recognition while preserving high computational efficiency. We have also shown that the performance of the state-of-the-art MBH descriptor can be improved with a discontinuity-preserving optical flow algorithm. In addition, a new method based on histogram intersection kernel was introduced to combine multiple channels of different descriptors. This method has the advantages of improving recognition accuracy with multiple descriptors and speeding up the classification process. On the efficiency side, we applied PCA to reduce the feature dimension which resulted in fast bag-of-features matching. We also evaluated the FLANN method on real-time action recognition. We conducted extensive experiments on real-world videos from challenging public action datasets. We showed that our methods achieved the state-of-the-art with real-time computational potential, thus highlighting the effectiveness and efficiency of the proposed methods.

The aim of this master thesis project was to develop methods for sample preparation and analysis by LC-MS suitable for global metabolomics of human plasma samples. In this thesis six different methods was tested, based on precipitation, solid phase extraction (SPE), and ultrafiltration. The methods differ in their mechanisms of action, but the end goal is the same, to remove the proteins and other high molecular weight compounds from the samples and to retain as many metabolites as possible in a reproducible manner. The LC-MS analysis were performed on a C18 and a HILIC type column using electrospray ionization (ESI) with both positive and negative ionization to cover as much of the metabolome as possible. The MarkerLynx software was then used to extract features from the chromatograms. The coefficient of variation (CV) was calculated and the features with CV &gt; 30 % were removed. The features were then used to compare the methods with each other in order to see whether any of the methods yielded the same capture of features. The largest amount of features was detected for precipitation with MeOH when using a C18 type column. For HILIC type columns precipitation with MeOH with a small addition of H3PO4 resulted in most unique features detected. The ionization mode was found to be less important, compared whit the choice of column, but more features was detected using positive mode than negative mode.

Šunų ausų ligos gana dažnai pasitaiko veterinarijos gydytojų praktikoje. Netiksliai diagnozavus ausies uždegimą gali būti pasirinktas netinkamas gydymo būdas, liga gali komplikuotis, sparčiai progresuoti, pasidaro lėtinė. Dažnai gyvūnai pradedami gydyti pasireiškus klinikiniams požymiams – vietiniam skausmui, niežėjimui, eritemai, epitelio pleiskanojimui ir susikaupus nemalonaus kvapo išskyroms. Ausų uždegimą gali sukelti labai daug priežasčių, kurios skirstomos į predisponuojančias, pirmines ir antrines. Predisponuojantiems veiksniams priskiriama šuns veislė, anatominė ausų stuktūra, jatrogeninės traumos, netinkamas gydymas, auglių obstrukcijos ar polypai. Infekciją gali sukelti ir į ausį patekę svetimkūniai, išoriniai dirgikliai: sieros sankaupos, kai ausys ilgai nevalomos, vanduo, patekęs į šuns ausis. Pirminiai veiksniai tiesiogiai sukelia išorinės ausies uždegimą. Tai ausų erkutės (Otodectes cynotis) ir odos parazitai (Demodex canis, Otodectes, Sarcoptes ir kt.), alergija, autoimuninės ligos, pirminė seborėja, blusos, hormonų pusiausvyros sutrikimas, endokrininiai sutrikimai. Antrinėms priežastims priskiriamos sąlyginai patogeninių bakterijų asociacijos ir grybeliai, kurie ausyje gyvena simbiozėje ir dauginasi. Bakterinį ausų uždegimą dažniausiai sukelia Staphylococcus spp., Pseudomonas aeruginosa, Proteus spp., rečiau – Bacillus ir Corynebacterium genčių mikroorganizmai bei Malassezia spp. mielės. Šio darbo tikslas buvo atlikti veiksnių, įtakojančių ausų uždegimo... [toliau žr. visą tekstą] / Ear diseases in dogs are common in veterinary medicine. Inaccurate diagnoses and inappropriate treatment of otitis, can rapidly cause serious progressive complications and the disease will be chronic. Frequently, animal treatments start when symptoms have been displayed: pain, head shaking, scratching at the pinnae, and malodorous ears, unpleasant smells and erythema are common complaints. Otitis can be caused by many reasons. There are distributive to predisposing, primary and secondary reasons. Predisposing factors are abnormal or breed-related conformation of the ear canal, including stenosis, yatrogenic injuries, inappropriate treatment hirsutism, and pendulous pinnae that restrict proper air flow into the ear canal. Excessive moisture, due to swimming or to frequent cleanings with improper solutions, can also lead to infections. Primary causes directly make inflammation of otitis externa. Parasites (Otodectes cynotis) and skin parasites (Demodex canis, Otodectes, Sarcoptes etc.), atopy, food allergy, auto-immune diseases primary seborrhea, fleas, and endocrine disorders. Secondary causes may be assigned to conditioned pathogenic bacteria as well as fungus which lives in ear with symbiosis and it is reproductive. Bacterial infections are often infected by Staphylococcus spp., Pseudomonas aeruginosa, Proteus spp., more rarely – Bacillus and Corynebacterium spp.and Malassezia spp. barm. The objective of the research work was to accomplish inflammation factors of otitis and... [to full text]

Darbo tikslas: dviem metodais gauti azodažiklius iš aromatinių aminų diazonio druskų, nustatyti jų pKa reikšmes, įvertinti spektrines savybes bei pritaikyti kaip indikatorius. Darbo uždaviniai: Gauti azodažiklius iš metilanilinų ir 4-nitroanilino vykdant jų diazonio druskų hidrolizę arba jungiant su 1-naftoliu bei 2-naftoliu, apskaičiuoti gautų junginių išeigas, nustatyti jų pKa reikšmes, spektrines savybes, galimą kaip indikatorių pritaikymą. Metodai: Tyrimui atlikti naudojami du azodažiklių sintezės metodai. Apskaičiuotos abiem metodais gautų azodažiklių procentinės išeigos. ESC įvertintas junginių grynumas. Gautų azodažiklių pKa reikšmės bei pH ribos, kuriose keičiasi jų spalva, vertinamos atliekant potenciometrinį titravimą. Vandeniniai azodažiklių tirpalai vertinti spektrofotometriškai, matuojant tirpalų absorbcijos maksimumus ir optinius tankius. Rezultatai: Azodažikliai, atliekant aromatinių aminų diazonio druskų hidrolizę (hidrolizatas sudarytas iš 44 ml vandens ir 20 ml koncentruotos H2SO4), gaunami veikiant aminus koncentruota HCl, kai hidrolizato temperatūra pakeliama iki 104-110 oC. Atliekant tradicinę azodažiklių sintezę išeigos yra geros-labai geros (vidutinė išeiga yra 76,95 proc.), o vykdant diazonio druskų hidrolizę – vidutinės (junginiai gaunami su vidutine 41,44 proc. išeiga). Tradiciniu metodu gautų naftolinių azodažiklių pKa reikšmės yra 8,8-11,0 ribose, o fenolinių, gautų atliekant eksperimentinę sintezę, - 7,5 8,0. Spektrofotometrinis tyrimas parodė... [toliau žr. visą tekstą] / Aim of the work: synthesis of azo dyes from aromatic amines diazonium salts by two methods, determination of their pKa values, spectral features and applicability as indicators of produced azo dyes. Main objectives: to obtain azo dyes from methylanilines and 4-nitroaniline by hydrolysis of their diazonium salts or by binding with 1-naphtol and 2-naphtol, to calculate yields of obtained compounds, to determine their pKa values, spectral features and their possible application as indicators. Methods: Two methods of azo dyes synthesis are used to perform the assay. Percentage yields of synthesized compounds, obtained by both methods, are calculated. Purity of compounds is evaluated by HPLC. pKa values and range of pH where the colour changes of received compounds are evaluated by potentiometric titration. Aqueous solutions of azo dyes are assessed by measuring absorption peaks and optical densities with a spectrophotometer. Results: it is necessary to synthesise aromatic amines diazonium salts with conc. HCl and to perform hydrolysis in 104-110 oC temperature when hydrolysate consist of 44 ml water and 20 ml conc. H2SO4 in order to obtain azo dyes by experimental method. Yields of traditional method are good-very good (average yield is 76,95 proc.), yields of experimental method are moderate (average yield is 41,44 proc.). pKa values of naphtolic azo dyes which are obtained by traditional method are 8,8-11,0 and phenolic azo dyes – 7,5-8,0. Spectrophotometric studies showed that... [to full text]

Near-infrared spectroscopy (NIRS) in an imaging technique that has gained much attention in brain-computer interfaces (BCIs). Previous NIRS-BCI studies have primarily employed temporal features, derived from the time course of hemodynamic activity, despite potential value contained in the spatial attributes of a response. In an initial offline study, we investigated the value of using joint spatial-temporal pattern classification with dynamic NIR topograms to differentiate intentional cortical activation from rest. With the inclusion of spatiotemporal features, we demonstrated a significant increase in achievable classification accuracies from those obtained using temporal features alone (p < 10-4). In a second study, we evaluated the feasibility of implementing joint spatial-temporal pattern classification in an online system. We developed an online system-paced NIRS-BCI, and were able to differentiate two cortical states with high accuracy (77.4±10.5%). Collectively, these findings demonstrate the value of including spatiotemporal features in the classification of functional NIRS data for BCI applications.