Speaker diarization: Segmentation and clustering of speeches

Speaker diarization is the process of sorting speeches according to the speaker. Diarization helps to search and retrieve what a certain speaker uttered in a meeting. Applications of diarization systemsextend to other domains than meetings, for example, lectures, telephone, television, and radio. Besides, diarization enhances the performance of several speech technologies such as speaker recognition, automatic transcription, and speaker tracking. Methodologies previously used in developing diarization systems are discussed. Prior results and techniques are studied and compared. Methods such as Hidden Markov Models and Gaussian Mixture Models that are used in speaker recognition and other speech technologies are also used in speaker diarization. The objective of this thesis is to develop a speaker diarization system in meeting domain. Experimental part of this work indicates that zero-crossing rate can be used effectively in breaking down the audio stream into segments, and adaptive Gaussian Models fit adequately short audio segments. Results show that 35 Gaussian Models and one second as average length of each segment are optimum values to build a diarization system for the tested data. Uniting the segments which are uttered by same speaker is done in a bottom-up clustering by a newapproach of categorizing the mixture weights.

Automatic image-based identification of Saimaa ringed seals

The Saimaa ringed seal is one of the most endangered seals in the world. It is a symbol of Lake Saimaa and a lot of effort have been applied to save it. Traditional methods of seal monitoring include capturing the animals and installing sensors on their bodies. These invasive methods for identifying can be painful and affect the behavior of the animals. Automatic identification of seals using computer vision provides a more humane method for the monitoring. This Master's thesis focuses on automatic image-based identification of the Saimaa ringed seals. This consists of detection and segmentation of a seal in an image, analysis of its ring patterns, and identification of the detected seal based on the features of the ring patterns. The proposed algorithm is evaluated with a dataset of 131 individual seals. Based on the experiments with 363 images, 81\% of the images were successfully segmented automatically. Furthermore, a new approach for interactive identification of Saimaa ringed seals is proposed. The results of this research are a starting point for future research in the topic of seal photo-identification.

Repeatability in automatic sorting of curved Norway spruce saw logs

Abstract

A Segmentation Model for Hi-tech Business Markets

Segmentointi on strateginen työkalu, joka tehostaa yrityksen resurssien käyttöä ja siten vaikuttaa kaikkiin asiakkuuksiin liittyviin liiketoimintaprosesseihin. Työn tavoitteena oli muodostaa segmentointimalli (sisältää sekä segmentointiprosessin että kriteerit) yritysinternetmarkkinoille. Työn tuloksia voidaan kuitenkin tulkita ja soveltaa laajemmin korkean teknologian yrityspalvelumarkkinoille. Tämä tutkielma lisää tietämystämme ja tarjoaa uudenlaisen näkemyksen segmentointiin korkean teknologian yrityspalvelumarkkinoilla. Työssä kuvataan korkean teknologian ja yritys- sekä palvelumarkkinoinnin erityispiirteitä ja kuinka nämä tekijät vaikuttavat segmentointimallin. Tutkimuksessa selvitettiin kohdeyrityksen nykyiset segmentointikäytännöt henkilökohtaisin asiantuntijahaastatteluin. Haastatteluiden avulla luotiin kuva nykyisistä lähestymistavoista sekä niiden lähtökohdista, vahvuuksista ja haasteista. Haastatteluiden analysoinnin jälkeen perustettiin projekti segmentoinnin kehittämiseksi. Työ tuloksena luotiin segmentointimalli, joka tarjoaa vankan perustan segmentoinnin kehittämiselle jatkuvana prosessina. Työssä esitetään segmentoinnin integroimista yrityksen asiakkuuksiin liittyviin liiketoimintaprosesseihin, joka usein puuttuu aiemmista töistä, sekä informaationkulun tehostamista segmentoinnin hyödyntämiseksi tehokkaammin. Segmentointi on strateginen työkalu ja vaatii siksi ylemmän johdon tuen ja sitoutumisen. Oikein sovellettuna segmentointi tarjoaa liiketoiminnalle mahdollisuuden merkittäviin etuihin kuten asiakastyytyväisyyden ja kannattavuuden kehittämiseen.

Segmentation model for small and medium size business sector: Case Oy Compaq Computer Ab

Markkinasegmentointi nousi esiin ensi kerran jo 50-luvulla ja se on ollut siitä lähtien yksi markkinoinnin peruskäsitteistä. Suuri osa segmentointia käsittelevästä tutkimuksesta on kuitenkin keskittynyt kuluttajamarkkinoiden segmentointiin yritys- ja teollisuusmarkkinoiden segmentoinnin jäädessä vähemmälle huomiolle. Tämän tutkimuksen tavoitteena on luoda segmentointimalli teollismarkkinoille tietotekniikan tuotteiden ja palveluiden tarjoajan näkökulmasta. Tarkoituksena on selvittää mahdollistavatko case-yrityksen nykyiset asiakastietokannat tehokkaan segmentoinnin, selvittää sopivat segmentointikriteerit sekä arvioida tulisiko tietokantoja kehittää ja kuinka niitä tulisi kehittää tehokkaamman segmentoinnin mahdollistamiseksi. Tarkoitus on luoda yksi malli eri liiketoimintayksiköille yhteisesti. Näin ollen eri yksiköiden tavoitteet tulee ottaa huomioon eturistiriitojen välttämiseksi. Tutkimusmetodologia on tapaustutkimus. Lähteinä tutkimuksessa käytettiin sekundäärisiä lähteitä sekä primäärejä lähteitä kuten case-yrityksen omia tietokantoja sekä haastatteluita. Tutkimuksen lähtökohtana oli tutkimusongelma: Voiko tietokantoihin perustuvaa segmentointia käyttää kannattavaan asiakassuhdejohtamiseen PK-yritys sektorilla? Tavoitteena on luoda segmentointimalli, joka hyödyntää tietokannoissa olevia tietoja tinkimättä kuitenkaan tehokkaan ja kannattavan segmentoinnin ehdoista. Teoriaosa tutkii segmentointia yleensä painottuen kuitenkin teolliseen markkinasegmentointiin. Tarkoituksena on luoda selkeä kuva erilaisista lähestymistavoista aiheeseen ja syventää näkemystä tärkeimpien teorioiden osalta. Tietokantojen analysointi osoitti selviä puutteita asiakastiedoissa. Peruskontaktitiedot löytyvät mutta segmentointia varten tietoa on erittäin rajoitetusti. Tietojen saantia jälleenmyyjiltä ja tukkureilta tulisi parantaa loppuasiakastietojen saannin takia. Segmentointi nykyisten tietojen varassa perustuu lähinnä sekundäärisiin tietoihin kuten toimialaan ja yrityskokoon. Näitäkään tietoja ei ole saatavilla kaikkien tietokannassa olevien yritysten kohdalta.

Automatic meter reading – benefits and applications

In liberalized electricity markets, which have taken place in many countries over the world, the electricity distribution companies operate in the competitive conditions. Therefore, accurate information about the customers’ energy consumption plays an essential role for the budget keeping of the distribution company and for correct planning and operation of the distribution network. This master’s thesis is focused on the description of the possible benefits for the electric utilities and residential customers from the automatic meter reading system usage. Major benefits of the AMR, illustrated in the thesis, are distribution network management, power quality monitoring, load modelling, and detection of the illegal usage of the electricity. By the example of the power system state estimation, it was illustrated that even the partial installation of the AMR in the customer side leads to more accurate data about the voltage and power levels in the whole network. The thesis also contains the description of the present situation of the AMR integration in Russia.

Analysis of Satellite Data for Forest Segmentation

In this work we study the classification of forest types using mathematics based image analysis on satellite data. We are interested in improving classification of forest segments when a combination of information from two or more different satellites is used. The experimental part is based on real satellite data originating from Canada. This thesis gives summary of the mathematics basics of the image analysis and supervised learning , methods that are used in the classification algorithm. Three data sets and four feature sets were investigated in this thesis. The considered feature sets were 1) histograms (quantiles) 2) variance 3) skewness and 4) kurtosis. Good overall performances were achieved when a combination of ASTERBAND and RADARSAT2 data sets was used.

Production Cycle Optimization: Defining the Market Potential for a New Planning Tool

An optimization tool has been developed to help companies to optimize their production cycles and thus improve their overall supply chain management processes. The application combines the functionality that traditional APS (Advanced Planning System) and ARP (Automatic Replenishment Program) systems provide into one optimization run. A qualitative study was organized to investigate opportunities to expand the product’s market base. Twelve personal interviews were conducted and the results were collected in industry specific production planning analyses. Five process industries were analyzed to identify the product’s suitability to each industry sector and the most important product development areas. Based on the research the paper and the plastic film industries remain the most potential industry sectors at this point. To be successful in other industry sectors some product enhancements would be required, including capabilities to optimize multiple sequential and parallel production cycles, handle sequencing of complex finishing operations and to include master planning capabilities to support overall supply chain optimization. In product sales and marketing processes the key to success is to find and reach the people who are involved directly with the problems that the optimization tool can help to solve.

Comparison Structure Analysis

This study presents an automatic, computer-aided analytical method called Comparison Structure Analysis (CSA), which can be applied to different dimensions of music. The aim of CSA is first and foremost practical: to produce dynamic and understandable representations of musical properties by evaluating the prevalence of a chosen musical data structure through a musical piece. Such a comparison structure may refer to a mathematical vector, a set, a matrix or another type of data structure and even a combination of data structures. CSA depends on an abstract systematic segmentation that allows for a statistical or mathematical survey of the data. To choose a comparison structure is to tune the apparatus to be sensitive to an exclusive set of musical properties. CSA settles somewhere between traditional music analysis and computer aided music information retrieval (MIR). Theoretically defined musical entities, such as pitch-class sets, set-classes and particular rhythm patterns are detected in compositions using pattern extraction and pattern comparison algorithms that are typical within the field of MIR. In principle, the idea of comparison structure analysis can be applied to any time-series type data and, in the music analytical context, to polyphonic as well as homophonic music. Tonal trends, set-class similarities, invertible counterpoints, voice-leading similarities, short-term modulations, rhythmic similarities and multiparametric changes in musical texture were studied. Since CSA allows for a highly accurate classification of compositions, its methods may be applicable to symbolic music information retrieval as well. The strength of CSA relies especially on the possibility to make comparisons between the observations concerning different musical parameters and to combine it with statistical and perhaps other music analytical methods. The results of CSA are dependent on the competence of the similarity measure. New similarity measures for tonal stability, rhythmic and set-class similarity measurements were proposed. The most advanced results were attained by employing the automated function generation – comparable with the so-called genetic programming – to search for an optimal model for set-class similarity measurements. However, the results of CSA seem to agree strongly, independent of the type of similarity function employed in the analysis.

Physical Realization of an Optical Automatic Target Recognition System

In the modern warfare there is an active development of a new trend connected with a robotic warfare. One of the critical elements of robotics warfare systems is an automatic target recognition system, allowing to recognize objects, based on the data received from sensors. This work considers aspects of optical realization of such a system by means of NIR target scanning at fixed wavelengths. An algorithm was designed, an experimental setup was built and samples of various modern gear and apparel materials were tested. For pattern testing the samples of actively arm engaged armies camouflages were chosen. Tests were performed both in clear atmosphere and in the artificial extremely humid and hot atmosphere to simulate field conditions.

Segmentation of Industrial Customers – Case: Finnish Mechanical Engineering Companies

Segmentointi on perinteisesti ollut erityisesti kuluttajamarkkinoinnin työkalu, mutta siirtymä tuotteista palveluihin on lisännyt segmentointitarvetta myös teollisilla markkinoilla. Tämän tutkimuksen tavoite on löytää selkeästi toisistaan erottuvia asiakasryhmiä suomalaisen liikkeenjohdon konsultointiyritys Synocus Groupin tarjoaman case-materiaalin pohjalta. K-means-klusteroinnin avulla löydetään kolme potentiaalista markkinasegmenttiä perustuen siihen, mitkä tarjoamaelementit 105 valikoitua suomalaisen kone- ja metallituoteteollisuuden asiakasta ovat maininneet tärkeimmiksi. Ensimmäinen klusteri on hintatietoiset asiakkaat, jotka laskevat yksikkökohtaisia hintoja. Toinen klusteri koostuu huolto-orientoituneista asiakkaista, jotka laskevat tuntikustannuksia ja maksimoivat konekannan käyttötunteja. Tälle kohderyhmälle kannattaisi ehkä markkinoida teknisiä palveluja ja huoltosopimuksia. Kolmas klusteri on tuottavuussuuntautuneet asiakkaat, jotka ovat kiinnostuneita suorituskyvyn kehittämisestä ja laskevat tonnikohtaisia kustannuksia. He tavoittelevat alempia kokonaiskustannuksia lisääntyneen suorituskyvyn, pidemmän käyttöiän ja alempien huoltokustannusten kautta.

Promoting Programming Learning. Engagement, Automatic Assessment with Immediate Feedback in Visualizations

The skill of programming is a key asset for every computer science student. Many studies have shown that this is a hard skill to learn and the outcomes of programming courses have often been substandard. Thus, a range of methods and tools have been developed to assist students’ learning processes. One of the biggest fields in computer science education is the use of visualizations as a learning aid and many visualization based tools have been developed to aid the learning process during last few decades. Studies conducted in this thesis focus on two different visualizationbased tools TRAKLA2 and ViLLE. This thesis includes results from multiple empirical studies about what kind of effects the introduction and usage of these tools have on students’ opinions and performance, and what kind of implications there are from a teacher’s point of view. The results from studies in this thesis show that students preferred to do web-based exercises, and felt that those exercises contributed to their learning. The usage of the tool motivated students to work harder during their course, which was shown in overall course performance and drop-out statistics. We have also shown that visualization-based tools can be used to enhance the learning process, and one of the key factors is the higher and active level of engagement (see. Engagement Taxonomy by Naps et al., 2002). The automatic grading accompanied with immediate feedback helps students to overcome obstacles during the learning process, and to grasp the key element in the learning task. These kinds of tools can help us to cope with the fact that many programming courses are overcrowded with limited teaching resources. These tools allows us to tackle this problem by utilizing automatic assessment in exercises that are most suitable to be done in the web (like tracing and simulation) since its supports students’ independent learning regardless of time and place. In summary, we can use our course’s resources more efficiently to increase the quality of the learning experience of the students and the teaching experience of the teacher, and even increase performance of the students. There are also methodological results from this thesis which contribute to developing insight into the conduct of empirical evaluations of new tools or techniques. When we evaluate a new tool, especially one accompanied with visualization, we need to give a proper introduction to it and to the graphical notation used by tool. The standard procedure should also include capturing the screen with audio to confirm that the participants of the experiment are doing what they are supposed to do. By taken such measures in the study of the learning impact of visualization support for learning, we can avoid drawing false conclusion from our experiments. As computer science educators, we face two important challenges. Firstly, we need to start to deliver the message in our own institution and all over the world about the new – scientifically proven – innovations in teaching like TRAKLA2 and ViLLE. Secondly, we have the relevant experience of conducting teaching related experiment, and thus we can support our colleagues to learn essential know-how of the research based improvement of their teaching. This change can transform academic teaching into publications and by utilizing this approach we can significantly increase the adoption of the new tools and techniques, and overall increase the knowledge of best-practices. In future, we need to combine our forces and tackle these universal and common problems together by creating multi-national and multiinstitutional research projects. We need to create a community and a platform in which we can share these best practices and at the same time conduct multi-national research projects easily.

Eye Fundus Image Analysis for Automatic Detection of Diabetic Retinopathy

Diabetes is a rapidly increasing worldwide problem which is characterised by defective metabolism of glucose that causes long-term dysfunction and failure of various organs. The most common complication of diabetes is diabetic retinopathy (DR), which is one of the primary causes of blindness and visual impairment in adults. The rapid increase of diabetes pushes the limits of the current DR screening capabilities for which the digital imaging of the eye fundus (retinal imaging), and automatic or semi-automatic image analysis algorithms provide a potential solution. In this work, the use of colour in the detection of diabetic retinopathy is statistically studied using a supervised algorithm based on one-class classification and Gaussian mixture model estimation. The presented algorithm distinguishes a certain diabetic lesion type from all other possible objects in eye fundus images by only estimating the probability density function of that certain lesion type. For the training and ground truth estimation, the algorithm combines manual annotations of several experts for which the best practices were experimentally selected. By assessing the algorithm’s performance while conducting experiments with the colour space selection, both illuminance and colour correction, and background class information, the use of colour in the detection of diabetic retinopathy was quantitatively evaluated. Another contribution of this work is the benchmarking framework for eye fundus image analysis algorithms needed for the development of the automatic DR detection algorithms. The benchmarking framework provides guidelines on how to construct a benchmarking database that comprises true patient images, ground truth, and an evaluation protocol. The evaluation is based on the standard receiver operating characteristics analysis and it follows the medical practice in the decision making providing protocols for image- and pixel-based evaluations. During the work, two public medical image databases with ground truth were published: DIARETDB0 and DIARETDB1. The framework, DR databases and the final algorithm, are made public in the web to set the baseline results for automatic detection of diabetic retinopathy. Although deviating from the general context of the thesis, a simple and effective optic disc localisation method is presented. The optic disc localisation is discussed, since normal eye fundus structures are fundamental in the characterisation of DR.