Šnekos signalų atpažinimo sistemų tikslumas priklauso nuo šnekos signalus aprašančių požymių ir šiuos požymius naudojančių klasifikatorių savybių. Vertinant tradiciškai atpažinimo sistemų tikslumą kiekvienai pasirinktai požymių sistemai ir kiekvienam klasifikatoriaus tipui tenka atlikti atpažinimo tikslumo skaičiavimus. Tokių darbų apimtis galima sumažinti įvertinus pasirenkamų požymių kokybę. Todėl buvo atlikti šnekos signalų požymių kokybės vertinimo tyrimai. Ištirtas metodas šnekos signalų atpažinimo požymių kokybei vertinti, grindžiamas trijų metrikų panaudojimu. Parodyta, kad tokiu būdu atrinkti šnekos signalų požymiai Euklido erdvėje aprašo atpažinimo sistemų kokybę ir leidžia sumažinti atpažinimo sistemų kokybės vertinimo darbų apimtis. Parodyta, kad šnekos signalų požymių kokybės vertinimo metodo algoritmo sudėtingumas yra O(2Rlog2R), o atpažinimo sistemos, kuriame naudojamas dinaminio laiko skalės kraipymo klasifikatorius, atpažinimo kokybės vertinimo algoritmo sudėtingumas yra O(R^2), R – šnekos signalų etalonų vektorių skaičius. Eksperimentinių tyrimų rezultatai patvirtino pateikto šnekos signalų atpažinimo požymių kokybės vertinimo metodo teisingumą. / The accuracy of speech recognition system depends on characteristics of employed speech recognition features and classifier. Evaluating the accuracy of speech recognition system in ordinary way, the error of speech recognition system has to be calculated for each type of explored feature system and each type of classifier. The amount of such calculations can be reduced if the quality of explored feature system is estimated. Accordingly, the researches were made for quality estimation of speech recognition features. The proposed method for quality estimation of speech recognition features is based on three metrics usage. It was demonstrated, that the proposed method describes the quality of speech recognition features in Euclidean space and reduces the calculations of quality estimation of speech recognition systems. Demonstrated, that algorithm complexity of method for quality estimation of speech recognition features is O(2Rlog2R), while algorithm complexity of dynamic time warping recognition system is O(R^2), where R is vectors number of speech pattern references. The results of experimental researches confirmed the correctness of the proposed method for quality estimation of speech recognition features.

The accuracy of speech recognition system depends on characteristics of employed speech recognition features and classifier. Evaluating the accuracy of speech recognition system in ordinary way, the error of speech recognition system has to be calculated for each type of explored feature system and each type of classifier. The amount of such calculations can be reduced if the quality of explored feature system is estimated. Accordingly, the researches were made for quality estimation of speech recognition features. The proposed method for quality estimation of speech recognition features is based on three metrics usage. It was demonstrated, that the proposed method describes the quality of speech recognition features in Euclidean space and reduces the calculations of quality estimation of speech recognition systems. Demonstrated, that algorithm complexity of method for quality estimation of speech recognition features is O(2Rlog2R), while algorithm complexity of dynamic time warping recognition system is O(R^2), where R is vectors number of speech pattern references. The results of experimental researches confirmed the correctness of the proposed method for quality estimation of speech recognition features. / Šnekos signalų atpažinimo sistemų tikslumas priklauso nuo šnekos signalus aprašančių požymių ir šiuos požymius naudojančių klasifikatorių savybių. Vertinant tradiciškai atpažinimo sistemų tikslumą kiekvienai pasirinktai požymių sistemai ir kiekvienam klasifikatoriaus tipui tenka atlikti atpažinimo tikslumo skaičiavimus. Tokių darbų apimtis galima sumažinti įvertinus pasirenkamų požymių kokybę. Todėl buvo atlikti šnekos signalų požymių kokybės vertinimo tyrimai. Ištirtas metodas šnekos signalų atpažinimo požymių kokybei vertinti, grindžiamas trijų metrikų panaudojimu. Parodyta, kad tokiu būdu atrinkti šnekos signalų požymiai Euklido erdvėje aprašo atpažinimo sistemų kokybę ir leidžia sumažinti atpažinimo sistemų kokybės vertinimo darbų apimtis. Parodyta, kad šnekos signalų požymių kokybės vertinimo metodo algoritmo sudėtingumas yra O(2Rlog2R), o atpažinimo sistemos, kuriame naudojamas dinaminio laiko skalės kraipymo klasifikatorius, atpažinimo kokybės vertinimo algoritmo sudėtingumas yra O(R^2), R – šnekos signalų etalonų vektorių skaičius. Eksperimentinių tyrimų rezultatai patvirtino pateikto šnekos signalų atpažinimo požymių kokybės vertinimo metodo teisingumą.

We hypothesized that temporal features of EEG are altered in sleep apnea subjects comparedto normal subjects. The initial aim was to develop a measure to discriminate sleep stages innormals. The longer-term goal was to apply these methods to identify differences in EEGactivity in sleep apnea subjects from normals. We analyzed the C3A2 EEG and anelectrooculogram (EOG) recorded from 9 normal adults awake and in rapid eye movement(REM) and non-REM sleep. The EEG signals were filtered to remove EOG contamination. Twomeasures of the irregularity of EEG signals, Sample Entropy (SpEn) and Tsallis Entropy, wereevaluated for their ability to discriminate sleep stages. SpEn changes with sleep state, beinglargest in Wake. Stage 3/4 had the smallest SpEn (0.57??0.11) normalized to Wake values,followed by Stage 2 (0.72??0.09), REM (0.75??0.1) and Stage 1 (0.89??0.05). This pattern wasconsistent in all the polysomnogram records analyzed. Similar pattern was observed in leadO1A2 as well. We conclude that SpEn may be useful as part of a montage for assessing sleepstate. We analyzed data from sleep apnea subjects having obstructive and central apnea eventsand have made some preliminary observations; the SpEn values were more similar across sleepstages and also high correlation with oxygen saturation was observed.

When a signal is decomposed into frequency bands, demodulated into modulator and carrier pairs, and portrayed in a carrier frequency-versus modulator frequency domain, significant information may be automatically observed about the signal. We refer to this domain as the modulation spectral domain. The modulation spectrum is referred to as a windowed Fourier transform across time that produces an acoustic frequency versus modulation frequency representation of a signal. Previously, frameworks incorporating the discrete short-time modulation transform (DSTMT) and modulation spectrum have been designed mostly for filtering of speech signals. This modulation spectral domain is rarely, if ever, discussed in typical signal processing courses today, and we believe its current associated tools and applications are somewhat limited. We seek to revisit this domain to uncover more intuition, develop new concepts to extend its capabilities, and increase its applications, especially in the area of music data mining. A recent interest has risen in using modulation spectral features, which are features in the modulation spectral domain, for music data mining. The field of music data mining, also known as music information retrieval (MIR), has been rapidly developing over the past decade or so. One reason for this development is the aim to develop frameworks leveraging the particular characteristics of music signals instead of simply copying methods previously applied to its speech-centered predecessors, such as speech recognition, speech synthesis, and speaker identification. This research seeks to broaden the perspective and use of an existing modulation filterbank framework by exploiting modulation features well suited for music signals. The objective of this thesis is to develop a framework for extracting modulation spectral features from music and other signals. The purpose of extracting features from these signals is to perform data mining tasks, such as unsupervised source identification, unsupervised source separation, and audio synthesis. More specifically, this research emphasizes the following: the usefulness of the DSTMT and the modulation spectrum for music data mining tasks; a new approach to unsupervised source identification using modulation spectral features; a new approach to unsupervised source separation; a newly introduced analysis of FM features in an AM-dominated modulation spectra; and other applications.

Darbo objektas. Antano Tatarės veikalo „Pamokslai išminties ir teisybės“ (Tatarė, 1987, Vilnius: Vaga, 20–143 psl.) zanavykų šnektos, priskiriamos vakarų aukštaičių kauniškių patarmei, tarminės ypatybės. Darbo hipotezė. A. Tatarės veikale „Pamokslai išminties ir teisybės“ atsispindi tik vakarų aukštaičių kauniškių pietinės šnektos – zanavykų – fonetinės ir morfologinės ypatybės. Darbo tikslas. Ištirti Antano Tatarės veikalo „Pamokslai išminties ir teisybės“ vakarų aukštaičių kauniškių zanavykų šnektos tarmines ypatybes. Darbo uždaviniai: 1. Naudojantis rastomis tarmybėmis, nurodyti pagrindines vakarų aukštaičių kauniškių zanavykų šnektos fonetines ir morfologines ypatybes; 2. Išrinktas vakarų aukštaičių patarmės zanavykų šnektos tarmines ypatybes suskirstyti pagal kalbos lygmenis (fonetikos, morfologijos, žodžių darybos, sintaksės, leksikos); 3. Naudojantis dialektologine, istorinės gramatikos ir kt. literatūra, ištirti rastus tarminius vartojimo atvejus. 4. Naudojantis rastais pavyzdžiais, apibūdinti A. Tatarės pastangas prisidedant prie kalbos norminimo darbų. Darbo metodai: 1. Aprašomasis-analitinis. Rastų fonetinių, morfologinių, sintaksinių, žodžių darybos, leksikos tarmybių aprašymas, remiantis fonetikos ir fonologijos, morfologijos, sintaksės, istorinės gramatikos, kalbos kultūros šaltiniais. 2. Lyginamasis. Tarmybių pavyzdžiai lyginami su bendrine kalba, artimoms zanavykams tarmėmis. Darbo struktūra. Darbą sudaro įvadas, dėstomoji dalis, išvados, priedai... [toliau žr. visą tekstą] / The object of the paper. The dialectal features of zanavykai dialect, labelled as the subdialect of western aukštaičiai kauniškiai, in Antanas Tatarė‘s work “Pamokslai išminties ir teisybės“ (Tatarė 1987, Vilnius: Vaga, 20–143 pages). The hypothesis of this paper. A. Tatarė‘s work “Pamokslai išminties ir teisybės“ reflects only phonetic and morphological features of the southern subdialect of western aukštaičiai kauniškiai (zanavykai). The aim of the paper is to explore the dialectal features of zanavykai dialect in Antanas Tatarė‘s work “Pamokslai išminties ir teisybės“. The goals of this paper are: 1. To indicate the main phonetic and morphological features of the subdialect of western aukštaičiai kauniškiai (zanavykai) by using found vernacularisms. To distribute the selected dialectal features of western aukštaičiai subdialect of zanavykai dialect according to language levels (phonetics, morphology, word formation, syntax, lexis); 2. To investigate the found cases of dialectal use by using dialectological, historical grammar and etc. literature. The methods of the paper: 1. The descriptive-analytic method. The description of found phonetic, morphological, syntactic, word formation, lexical vernacularisms, based on the resources of phonetics and phonology, morphology, syntax, historical grammar, language culture. 2. The comparative method. The examples of vernacularisms are compared with common language, dialects close to zanavykai. The stucture of this paper. The work... [to full text]

Gastrointestinės stromos tumorai (GIST) – dažniausi mezenchiminės kilmės virškinamojo trakto navikai, kurie pradėti identifikuoti ir diagnozuoti tik per pastaruosius 25 m. GIST, pagal ligos progresavimo reliatyvią riziką, klasifikuojami į rizikos kategorijas, kurių nustatymas 2002 m. Nacionalinių sveikatos institutų sutarimu (NSI) paremtas dviem kintamaisiais: naviko dydžiu ir navikinių ląstelių mitozių skaičiumi 50 DPRL, kurių apimamas plotas įvairių studijų ir rekomendacijų duomenimis varijuoja nuo 5 mm² iki 11,9 mm². Neaišku, kaip netikslus mitozių skaičius įtakoja rizikos kategorijos nustatymą bei jų sąsajas su ligos progresavimu. Nors atliktose didžiausiose GIST studijose teigiama, kad, esant identiškiems morfologiniams požymiams, skrandžio GIST prognozė geresnė, kiti tyrėjai nustatė morfologinius požymius lemiančius geresnę šių navikų prognozę. Esant prieštaringoms nuomonėms, tyrimo metu nustatėme ir palyginome skirtingų lokalizacijų GIST morfologinius požymius ir prognozę. NSI klasifikacijoje buvo pastebėta, kad ligos prognozę įtakoja ir GIST histologinis fenotipas, tačiau tuomet nebuvo pakankamai atliktų studijų, paneigiančių ar patvirtinančių šiuos teiginius, todėl detaliai nagrinėjome skirtingų histologinių fenotipų GIST morfologinius požymius ir jų įtaką ligos prognozei bei kitų morfologinių požymių: naviko dydžio, mitozių skaičiaus, ląstelingumo, branduolių polimorfizmo ir nekrozės sąsajas su ligos prognoze. / Gastrointestinal stromal tumours (GIST) – the most common mesenchymal tumours of the gastrointestinal tract which have been identified and diagnosed in the past 25 years. GIST, according to relative risk of disease progression, are classified in risk categories which under the agreement of the National Institutes of Health (2002) were based on two variables: tumour size and mitotic count in 50 HPF, which total area varies from 5 mm² to 11.9 mm². There is no consensus what area is sufficient for the counting of mitoses. Thus, we compare number of GIST tumor cell mitoses in different sized areas and evaluated the impact of mitotic rate changes in determination of risk categories and their relation to the progression of the disease. According to the data of largest GIST study, disease prognosis is also influenced by tumour localization, other researchers found morphological features of these tumours leading to a better prognosis. Thus, we identified and compared the different localizations of GIST morphological features and prognosis. It was observed that disease prognosis is also determined by the tumour histological phenotype, but up till now the controversial results of histological phenotypes relationship to disease prognosis are published. We set morphological features of different GIST histological phenotypes and evaluate their relation to disease progression, evaluate the impact of GIST morphological features to disease prognosis.

The usage of 3D-modeling is expanding rapidly. Modeling from aerial imagery has become very popular due to its increasing number of both civilian and mili- tary applications like urban planning, navigation and target acquisition. This master thesis project was carried out at Vricon Systems at SAAB. The Vricon system produces high resolution geospatial 3D data based on aerial imagery from manned aircrafts, unmanned aerial vehicles (UAV) and satellites. The aim of this work was to investigate to what degree superpixel segmentation and supervised learning can be applied to a terrain classification problem using imagery and digital surface models (dsm). The aim was also to investigate how the height information from the digital surface model may contribute compared to the information from the grayscale values. The goal was to identify buildings, trees and ground. Another task was to evaluate existing methods, and compare results. The approach for solving the stated goal was divided into several parts. The first part was to segment the image using superpixel segmentation, after that features were extracted. Then the classifiers were created and trained and finally the classifiers were evaluated. The classification method that obtained the best results in this thesis had approx- imately 90 % correctly labeled superpixels. The result was equal, if not better, compared to other solutions available on the market.

This thesis examines the occurrence of hydrologic variability in subarctic ponds within the Churchill region of the Hudson Bay Lowland (HBL) and investigates the utility of using remote sensing studies to characterize changes in pond surface area. The thesis also characterizes hydro-climatic change over the past ~60 years, and compares this to pond sustainability within the region of Churchill. A multiple-methods approach incorporating field research, simple water balance modeling and remote sensing is used to address these objectives. Research findings demonstrate the occurrence of natural fluctuations in pond surface area and water levels in the Canadian subarctic. These fluctuations in pond water levels (and thus surface area) are caused by differences in antecedent hydrologic conditions, which are easily detected using remotely sensed imagery and may produce unrepresentative estimates of pond surface area change. Resulting from a 4.5 - 11.8 cm variation in water depth, pond surface areas were significantly altered by antecedent precipitation (average: 3,711 m²), intra-seasonal variability (average: 2,049 m²) and inter-annual climatic variations (average: 1,977 m²). These noteworthy pond boundary and water level differences reinforce the importance of accounting for hydrologic variability when delineating representative pond coverage and sustainability. Contemporary pond sustainability findings reveal significant regional climatic change, changing pond hydrologic conditions and overall pond physical stability between 1947 and 2008. Specifically, the Churchill region has become warmer and wetter. Occurring at a rate of 1.37 mm/yr over the study period, changing atmospheric conditions caused a decrease in open water pond hydrologic deficits. During the hydrologic recharge period, modeled pond water levels exhibited an increasing trend (August +0.72 mm/yr, September 0.51 mm/yr), which suggests ponds are filling closer to their maximum storage capacity prior to freeze-up. A remote sensing analysis of pond boundary modifications in mid-summer revealed no change in contemporary physical pond sustainability. Detected surface area changes from imagery were mainly attributed to naturally induced hydrologic variability. Overall, this thesis suggests a new methodological approach for conducting remote sensing pond sustainability research within the arctic/subarctic environment. As well, this study determined pond sustainability within the Churchill region over the last ~60 years.

Glaciers and ice sheets covered extensive areas in the Northern Hemisphere during the last glacial period. Subsequently to the Last Glacial Maximum (LGM), they retreated rapidly and, except for Greenland and some other ice caps and glaciers, they vanished after the last glacial termination. This thesis examines the dynamics of deglacial environments by analysing the glacial geomorphological record with focus on the landforms created by glacial meltwater. The aims are (i) to evaluate the data available for mapping glacial meltwater features at the regional scale, and (ii) to demonstrate the potential of such features for regional ice retreat reconstructions in high-relief landscapes. Meltwater landforms such as ice-marginal meltwater channels, eskers, deltas and fossil glacial lake shorelines are used to infer former ice surface slope directions and successive positions of retreating ice margins. Evaluated high-resolution satellite imagery and digital elevation models reveal their potential to replace aerial photographs as the primary data for mapping glacial meltwater landforms. Following a methods study, reconstructions of the deglacial dynamics are carried out for central Transbaikalia, Siberia, Russia, and for the Cordilleran Ice Sheet (CIS) in central British Columbia, Canada, using regional geomorphological mapping surveys. Mapped glacial landforms in central Transbaikalia show evidence of a significant glaciation that possibly extended beyond the high mountain areas. Large glacial lakes were formed as advancing glaciers blocked rivers, and of these, Glacial Lake Vitim was the most prominent. Deglacial dynamics of the CIS reveals that the ice divide shifted to the Coast Mountains in north-central British Columbia and the eastern ice margin retreated towards the ice divide in late glacial time. This thesis demonstrates the potential to reconstruct ice retreat patterns and deglacial dynamics at regional scales by interpretation of the meltwater landform record. / At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 5: Submitted. Paper 6: Manuscript.

Object tracking is an active research topic in computer vision and has appli- cation in several areas, such as event detection and robotics. Vehicle tracking is used in Intelligent Transport System (ITS) and surveillance systems. Its re- liability is critical to the overall performance of these systems. Feature-based methods that are used to represent distinctive content in visual frames are one approach to vehicle tracking. Existing feature-based tracking systems can only track vehicles under ideal conditions. They have difficulties when used under a variety of conditions, for example, during both the day and night. They are highly dependent on stable local features that can be tracked for a long time period. These local features are easily lost because of their local property and image noise caused by factors such as, headlight reflections and sun glare. This thesis presents a new approach, addressing the reliability issues mentioned above, tracking whole feature groups composed of feature points extracted with the Scale Invariant Feature Transform (SIFT) algorithm. A feature group in- cludes several features that share a similar property over a time period and can be tracked to the next frame by tracking individual feature points inside it. It is lost only when all of the features in it are lost in the next frame. We cre- ate these feature groups by clustering individual feature points using distance, velocity and acceleration information between two consecutive frames. These feature groups are then hierarchically clustered by their inter-group distance, velocity and acceleration information. Experimental results show that the pro- posed vehicle tracking system can track vehicles with the average accuracy of over 95%, even when the vehicles have complex motions in noisy scenes. It gen- erally works well even in difficult environments, such as for rainy days, windy days, and at night. We are surprised to find that our tracking system locates and tracks motor bikes and pedestrians. This could open up wider opportunities and further investigation and experiments are required to confirm the tracking performance for these objects. Further work is also required to track more com- plex motions, such as rotation and articulated objects with different motions on different parts.