Search

School bullying is a common social problem among teenagers that occurs in various ways, and physical violence is considered the most harmful. However, all the existing methods for reporting school violence are human-driven, making them inconvenient for victims to use. The author’s research group therefore aims to develop automatic violence recognition methods. First, this thesis proposed a Fuzzy Multi-Threshold algorithm. However, as more activities involving players of different ages and genders were added, it failed to perform effectively. A Proportional K-Nearest Neighbors algorithm was therefore proposed to address this issue, achieving an average recognition accuracy of 80%. Speech emotion recognition was then incorporated. Time- and frequency-domain motion features were extracted for activity recognition, while Mel Frequency Cepstrum Coefficients were extracted for emotion recognition. A Back Propagation Neural Network was used as the classifier, achieving an average accuracy of 87.8%. To further enhance recognition performance, an additional movement sensor was utilized. Motion features were separately extracted from each sensor. An improved Relief-F algorithm was proposed to select useful and low-redundancy motion features, while a Radial Basis Function neural network was used as the classifier. An improved Dempster-Shafer fusion algorithm was proposed to combine the classification results of each sensor. An average recognition accuracy of 93.6% was achieved. To address the problem of bullies removing wearable sensors, an image-based violence recognition was proposed. Surveillance cameras capture images, and target-detecting algorithms and morphological processing methods were utilized to detect foreground targets. A circumscribed rectangular frame (CRF) integration method was proposed to refine the detected targets. CRF features and optical flow features were extracted, and three different selection methods were employed to select useful features., Tiivistelmä Koulukiusaaminen on teini-ikäisten yleinen sosiaalinen ongelma, ja sitä tapahtuu monin tavoin fyysisen väkivallan ollessa vahingollisin. Nykyiset kouluväkivallasta ilmoittamisen menetelmät ovat kuitenkin uhrin aloitteeseen perustuvia ja näin ollen epämukavia käyttää. Tässä tutkimustyössä kehitetään kouluväkivallan tunnistamismenetelmiä, jotka havaitsevat kiusaamistapahtumat automaattisesti. Työssä kehitettiin ensin Fuzzy Multi-Threshold -algoritmi, joka kykeni luokittelemaan seitsemän erityyppistä fyysistä aktiviteettia. Ratkaisu kuitenkin menetti tarkkuuttaan, kun aineistoa laajennettiin eri-ikäisillä ja erisukupuolisilla koehenkilöillä. Seuraavaksi kehitettiin Proportional K-Nearest Neighbors -algoritmi ja tuloksena oli 80 % tunnistamistarkkuus. Ratkaisuun lisättiin emootion tunnistus puhesignaalista. Fyysisen aktiivisuuden piirteet laskettiin liikesignaalin aika- ja taajuusalueista ja puheäänen Mel Frequency Cepstrum Coefficients -piirteet laskettiin tunteiden tunnistamiseksi. Back Propagation -neuroverkko antoi tunnistustarkkuudeksi 87,8 %. Ratkaisuun lisättiin toinenkin liikeanturi ja piirteet laskettiin molemmista anturisignaaleista. Työssä kehitettiin parannettu Relief-F-algoritmi parhaiden piirteiden valintaan liikkeen ominaisuuksista. Luokittelijana käytettiin Radial Basis Function -neuroverkkoa. Anturien tuottamien signaalien luokittelutulosten yhdistäminen tehtiin Dempster-Shafer-fuusioalgoritmilla, jolloin saavutettiin 93,6 % luokittelutarkkuus. Koska on mahdollista, että kiusaajat poistavat puettavat anturit tai älypuhelimet, kehitettiin vaihtoehtoinen tapa tunnistaa kouluväkivalta videokuvien perusteella. Valvontakameran kuvan edustassa sijaitsevien kohteiden paikantamiseksi käytettiin kohteen havaitsemisalgoritmeja ja morfologisia operaatioita. Kohteen ympäröivälle suorakaiteelle laskettiin kuvapiirteet sisältäen optisen virtauksen piirteet, ja niistä valittiin parhaat kolmella eri piirteenvalintamenetelmällä. 2-kerroksinen luokittelija De

Environmental and natural resource issues are interconnected and interdependent. Understanding these challenges and mitigating their consequences calls for interdisciplinary research and research-based decision-making. In Finland, a significant part of the research that supports decision-making and policy processes is conducted within administrative sectors and is known as sectoral research. However, a growing number of research needs and practices are horizontal, i.e. cross-sectoral. This thesis examines the promotion of a horizontal approach to sectoral research in the field of environment and natural resources. An interdisciplinary thesis utilises open systems theory, systems thinking, evolutionary organisational theory and complexity thinking, and uses concepts derived from biology in the manner of metaphor to interpret phenomena and processes. The first part of the thesis argues how Finland’s sectoral research system evolved from the 1990s to the 2010s. This change aimed to adapt the system to better tackle society’s forthcoming horizontal challenges. The second sub-study delves into the critical factors related to the acquisition of sectoral research. The particular focus is on the concept of commissioning competence. The third part of the thesis explores the potential for collaboration in addressing complex issues. This research analyses the research programmes as elements and dynamics that establish interactions within the research system. The thesis demonstrates that efforts were undertaken to facilitate structural changes within the research system through the strengthening of market mechanisms and networking. However, the prioritisation issues between horizontal and sectoral research needs created friction and, therefore, the desired self-organisation was not accomplished. The long-term analysis also reveals the co-evolutionary developments in which research policy goals and research institution practices underwent changes in interaction with one, Tiivistelmä Ympäristö- ja luonnonvarakysymyksien ymmärtämiseksi ja niiden vaikutuksiin varautumiseksi tarvitaan monitieteistä tutkimusta sekä tieteeseen perustuvaa yhteiskunnallista päätöksentekoa. Suomessa päätöksentekoa ja politiikan prosesseja tukevaa tukimusta tehdään merkittäviltä osin hallinnollisten sektoreiden puitteissa sektoritutkimuksena. Tutkimuksen tarpeet ja toimintatavat ovat kuitenkin yhä suuremmalta osalta horisontaalisia eli sektorirajat ylittäviä. Tässä väitöstutkimuksessa tarkastellaan horisontaalisen lähestymistavan edistämistä ympäristö- ja luonnonvara-alan sektoritutkimuksessa. Monitieteisessä väitöstutkimuksessa hyödynnetään systeemiteoriaa, systeemiajattelua, evolutiivista organisaatioteoriaa ja kompleksisuusajattelua sekä käytetään biologiasta metaforan tapaan johdettuja käsitteitä ilmiöiden ja tapahtumien tulkitsemiseen. Ensimmäisessä osatyössä tulkitaan tutkimusjärjestelmän rakenteellista muutosta evolutionäärisenä kehityskulkuna. Tutkimuksessa tarkastellaan ajanjaksoa 1990-luvulta 2010-luvulle. Toisessa osatyössä syvennetään tutkimuksen tilaajaosaamisen käsitettä. Työssä kehitetään konseptuaalinen malli, joka havainnollistaa tilaajaosaamisen moniulotteista luonnetta ja perustelee samalla sen merkitystä ja roolia tutkimusjärjestelmän rakenteellisessa kehittämisessä. Kolmannessa osatyössä analysoidaan yhteistyörakenteiden ja tutkimusohjelmien mahdollisuuksia kompleksisten aiheiden hallinnassa. Väitöstutkimus osoittaa, että tutkimusjärjestelmän rakenteellista muutosta pyrittiin mahdollistamaan markkinamekanismia ja verkostoitumista vahvistamalla. Ristiriidat horisontaalisten ja sektoraalisten tutkimustarpeiden priorisoinnissa aiheuttivat muutoskitkaa. Pitkänajan analyysi tuo esiin myös yhteisevoluutiomaisia kehityskulkuja, joissa tutkimuspolitiikan tavoitteet ja tutkimuslaitosten käytänteet muuttuivat keskinäisessä vuorovaikutuksessa. Hallinnonalojen tutkimustiedon hankinnassa ja tutkimuslaitosten ohjauksessa merkityksellinen tilaajaosaamin

Campus violence is a common social phenomenon all over the world, and is the most harmful type of school bullying events. As artificial intelligence and remote sensing techniques develop, there are several possible methods to detect campus violence, e.g., movement sensor-based methods and video sequence-based methods. Sensors and surveillance cameras are used to detect campus violence. In this paper, the authors use image features and acoustic features for campus violence detection. Campus violence data are gathered by role-playing, and 4096-dimension feature vectors are extracted from every 16 frames of video images. The C3D (Convolutional 3D) neural network is used for feature extraction and classification, and an average recognition accuracy of 92.00% is achieved. Mel-frequency cepstral coefficients (MFCCs) are extracted as acoustic features, and three speech emotion databases are involved. The C3D neural network is used for classification, and the average recognition accuracies are 88.33%, 95.00%, and 91.67%, respectively. To solve the problem of evidence conflict, the authors propose an improved Dempster–Shafer (D–S) algorithm. Compared with existing D–S theory, the improved algorithm increases the recognition accuracy by 10.79%, and the recognition accuracy can ultimately reach 97.00%.

In daily life, speech is the main medium of human communication, and interpersonal communication is emotional. People hope that the computer can give a response based on the emotions contained in the voice. In this paper, we build a Wechat program of speech emotion recognition system, which is based on a random forest classifier. Firstly, the system preprocesses the collected speech signals in order to reduce noise. Secondly, 16 acoustic features are extracted from the pre-processed speech signals. The system obtains the emotional features of speech by applying 12 statistical functions to the original acoustic features. The emotional classification of Berlin Speech Emotion Database uses two classifiers: the Random Forest Classifier and the Support Vector Machine. The recognition accuracy of the SVM classifier is 83%. The accuracy of the random forest classifier is 89%. Finally, the random forest classifier is used to build the speech emotion recognition system.

Violence is a serious global problem that may cause considerable distress with long lasting consequences. It affects all associated parties, including victims, perpetrators and witnesses. Studies on automatic violence detection using surveillance cameras have been conducted for about 20 years, but they are limited to the witness perspective. This work aimed to approach the problem from the victims’ point of view. The main research question was: is it possible to classify events as violent or non-violent by analysing ultra-short-term ECG signals? Two specific objectives were set: 1) to collect a small-scale database containing synchronized ultra-short-term ECG signals from both physically violent and non-violent events based on simulations, and 2) to extract features from ECG signals to discriminate between violent and non-violent events indirectly and directly. In the indirect inference approach, emotion recognition served as a first step to separating between violent and non-violent events; this work was done using the Mahnob-HCI database. The direct inference approach, in contrast, classified signals as being indicative of either violent or non-violent events. The main contributions are as follows: 1) For the indirect approach, methods were developed to extract useful features. These were mainly based on bivariate empirical mode decomposition, performed to obtain intrinsic mode functions, which were examined by spectrogram analysis to reveal dominant frequencies and by statistical analysis to expose oscillation characteristics. Enhancing these features with supervised dimensionality reduction achieved an accuracy of up to 70% for a 3-class problem in valence and arousal. 2) For the direct approach with the collected database, the same feature extraction method was applied to achieve an accuracy of up to 88% for a binary classification problem of discriminating between violent and non-violent events. Enhancing features with supervised dimensionality reduct, Tiivistelmä Väkivalta on vakava globaali ongelma, joka saattaa aiheuttaa kohteille huomattavaa kärsimystä ja pitkäaikaisia seuraamuksia. Se vaikuttaa kaikkiin osapuoliin: uhreihin, tekijöihin ja todistajiin. Automaattista väkivallan ilmaisua on tutkittu turvakameroita käyttäen noin 20 vuotta, mutta se on rajoittunut todistajan näkökulmaan. Tässä työssä pyrittiin lähestymään ongelmaa uhrin näkökulmasta. Tärkein tutkimuskysymys oli: Onko mahdollista luokitella tapahtumia väkivaltaisiksi ja ei-väkivaltaisiksi analysoimalla ultralyhyitä EKG-näytteitä. Työssä asetettiin kaksi erityistavoitetta: 1) Kerätään pienen mittakaavan tietokanta, joka sisältää synkronoitua ultralyhyttä EKG-signaalia simuloiduista väkivaltaisista ja ei-väkivaltaisista tapahtumista, 2) irrotetaan EKG-signaalista piirteitä, joiden avulla väkivaltaisia ja ei-väkivaltaisia tapahtumia voidaan erotella toisistaan suoraan ja epäsuorasti. Epäsuorassa päättelyssä tunnetilojen tunnistaminen toimi ensiaskeleena väkivaltaisten ja ei-väkivaltaisten tapahtumien erottelussa; tutkimus tehtiin Mahnob-HCI-tietokantaa käyttäen. Suora päättely sen sijaan luokitteli signaalit suoraan väkivaltaisiin ja ei-väkivaltaisiin tapahtumiin liittyviksi. Tärkeimmät kontribuutiot ovat seuraavat: 1) Epäsuoraa lähestymistapaa varten kehitettiin menetelmiä, joilla irrotetaan luokittelijalle hyödyllisiä piirteitä. Ne soveltavat pääosin bivariate empirical mode decomposition -menetelmää irrottamaan ultralyhyistä EKG-näytteistä muotofunktioita, joiden dominanttitaajuuksia tutkittiin spektrianalyysillä. Luokittelijan avainpiirteinä olivat dominoivien taajuuksien ja niiden ensimmäisten derivaattojen tilastolliset jakaumat. Tehostamalla näitä piirteitä ohjatulla dimensionaalisuuden vähentämisellä saatiin aikaan jopa 70 %:n tarkkuus valenssin ja arousalin 3-luokkaisessa asetelmassa. 2) Suorassa lähestymistavassa käytettiin samaa piirteenirrotusmenetelmää, jolla saatiin aikaan jopa 88 %:n tarkkuus väkivalta- ja ei-väkivaltatapahtumien erotte

Wearable devices provide their users with feedback and guidance for daily behavior. The objective of this thesis was to develop new algorithms for wearable devices and validate them for long-term assessment of human health-related behavior. In addition to technical accuracy, attention was also paid to feasible long-term use and usability as a research tool at big scale. The main sensor arrangements in this work were a digital accelerometer integrated into a wrist-band and optical photoplethysmographic sensor integrated into a wellness ring. In this work, new algorithms were developed for quantification of daily physical activity and sleep schedule. The same technology was also used to evaluate physiological responses in resting heart rate and heart rate variability. The measurements were validated against gold standard methods. In addition to separate measurements, the use of the sensor solutions was extended to a combined big data evaluation of lifestyle factors, sleep quality and body responses. The algorithms embedded in the wrist-band were able to detect a training-induced change in daily energy expenditure. The developed models explained 62–85% of the inter-individual variability in total energy expenditure. High agreement between ring PPG and the ECG based measurement of night-time heart rate variability was shown (r2=0.980). In addition, the results showed that the data gathered with the ring reflect behavioral, societal and seasonal effects on resting heart rate at the population scale. Moreover, the collected data displayed how bedtime inconsistency and alcohol consumption negatively affect human recovery by means of reduced HRV and worsened sleep quality. The solutions developed in this work combine daily physical activity and nightly recovery metrics. The methodology has already been embedded in commercial devices that are used world-wide. The biggest dataset of this work analyzes human body responses from over 55,000 subjects and provides the hi, Tiivistelmä Puettavan teknologian tuotteet tarjoavat käyttäjilleen henkilökohtaisen palautejärjestelmän päivittäisen käyttäytymisen seurantaan ja ohjaukseen. Tämän tutkimuksen tavoitteena oli kehittää ja validoida uutta puettavaa anturiteknologiaa terveyskäyttäytymisen mittaamiseen. Teknisen suorituskyvyn lisäksi kehitettävältä teknologialta edellytettiin käyttömukavuutta pitkäaikaisissa mittauksissa ja käyttökelpoisuutta tutkimustyökaluna laajoissa aineistoissa. Väitöskirjatyön tärkeimmät mittausjärjestelyt olivat rannekkeeseen integroitu digitaalinen kiihtyvyysanturi ja hyvinvointisormukseen integroitu pulssimittaus. Työssä kehitettiin laskentamenetelmät päivittäisen liikunta-aktiivisuuden ja unirytmin mittausta varten. Samalla teknologialla mitattiin myös kehon fysiologisia vasteita — leposykettä ja sykevaihtelua. Mittaukset validoitiin luotettavilla vertailumenetelmillä. Erillisten mittausten lisäksi työssä yhdistettiin päivän toiminnan, yöunen laadun ja kehon vasteiden havainnointi laajoissa aineistoissa. Rannemittariin integroidulla laskentamenetelmällä pystyttiin toteamaan liikuntaharjoittelun vaikutus päivittäiseen energiankulutukseen. Kehitetyt mallit selittivät 62–85 % yksilöiden välisestä päivittäisen energiankulutuksen vaihtelusta. Sormuksessa toimiva sykevaihtelun arviointimenetelmä voitiin todeta erittäin luotettavaksi EKG:hen verrattuna yön yli tehtävissä mittauksissa (r2 = 0,980). Tulosten mukaan sormuksella kerätyllä datalla pystyttiin myös havainnollistamaan käyttäytymisen ja vuodenaikoihin liittyvien tekijöiden vaikutuksia leposykkeeseen ja sykevaihteluun. Analysoitu data esimerkiksi osoitti unirytmin epäsäännöllisyyden ja alkoholinkäytön vaikuttavan kielteisesti palautumiseen, mikä ilmeni alentuneena sykevaihteluna ja huonompana unen laatuna. Työssä kehitetyt ratkaisut yhdistävät päivän aktiivisuuden ja yön palauttavuuden mittauksen. Menetelmät on otettu käyttöön kuluttajatuotteissa, joita käytetään eri puolilla maailmaa. Suurin aineisto tarkaste

School bullying is a serious problem among teenagers. School violence is one type of school bullying and considered to be the most harmful. As AI (Artificial Intelligence) techniques develop, there are now new methods to detect school violence. This paper proposes a video-based school violence detecting algorithm. This algorithm first detects foreground moving targets via the KNN (K-Nearest Neighbor) method and then preprocesses the detected targets via morphological processing methods. Then, this paper proposes a circumscribed rectangular frame integrating method to optimize the circumscribed rectangular frame of moving targets. Rectangular frame features and optical-flow features were extracted to describe the differences between school violence and daily-life activities. We used the Relief-F and Wrapper algorithms to reduce the feature dimension. SVM (Support Vector Machine) was applied as the classifier, and 5-fold cross validation was performed. The accuracy was 89.6%, and the precision was 94.4%. To further improve the recognition performance, we developed a DT–SVM (Decision Tree–SVM) two-layer classifier. We used boxplots to determine some features of the DT layer that are able to distinguish between typical physical violence and daily-life activities and between typical daily-life activities and physical violence. For the remainder of activities, the SVM layer performed a classification. For this DT–SVM classifier, the accuracy reached 97.6%, and the precision reached 97.2%, thus showing a significant improvement.

Feature fusion is a common approach to improve the accuracy of the system. Several attemps have been made using this approach on the Mahnob-HCI database for affective recognition, achieving 76% and 68% for valence and arousal respectively as the highest achievements. This study aimed to improve the baselines for both valence and arousal using feature fusion of HRV-based, which used the standard Heart Rate Variability analysis, standardized to mean/standard deviation and normalized to [-1,1], and cvxEDA-based feature, calculated based on a convex optimization approach, to get the new baselines for this database. The selected features, after applying the sequential forward floating search (SFFS), were enhanced by the Neighborhood Component Analysis and fed to kNN classifier to solve 3-class classification problem, validated using leave-one-out (LOO), leave-one-subject-out (LOSO), and 10-fold cross validation methods. The standardized HRV-based features were not selected during the SFFS method, leaving feature fusion from normalized HRV-based and cvxEDA-based features only. The results were compared to previous studies using both single- and multi-modality. Applying the NCA enhanced the features such that the performances in valence set new baselines: 82.4% (LOO validation), 79.6% (10-fold cross validation), and 81.9% (LOSO validation), enhanced the best achievement from both single- and multi-modality. For arousal, the performances were 78.3%, 78.7%, and 77.7% for LOO, LOSO, and 10-fold cross validations respectively. They outperformed the best achievement using feature fusion but could not enhance the performance in single-modality study using cvxEDA-based feature. Some future works include utilizing other feature extraction methods and using more sophisticated classifier other than the simple kNN.

School violence is a serious problem all over the world, and violence detection is significant to protect juveniles. School violence can be detected from the biological signals of victims, and emotion recognition is an important way to detect violence events. In this research, a violence simulation experiment was designed and performed for school violence detection system. Emotional voice from the experiment was extracted and analyzed. Consecutive elimination process (CEP) algorithm was proposed for emotion recognition in this paper. After parameters optimization, SVM was chosen as the classifier and the algorithm was validated by Berlin database which is an emotional speech database of adults, and the mean accuracy for seven emotions was 79.05%. The emotional speech database of children extracted in violence simulation was also classified by SVM classifier with proposed CEP algorithm, and the mean accuracy was 66.13%. The results showed that high classification performance could be achieved with the CEP algorithm. The classification result was also compared with database of adults, and the results indicated that children and adults’ voice should be treated differently in speech emotion recognition researches. The accuracy of children database is lower than adult database; the accuracy of violence detection will be improved by other signals in the system.

The eutrophication of the Baltic Sea continues despite decrease of the external phosphorus load by as much as 80% of the target confirmed by HELCOM. The aim of this thesis is to investigate this paradox, critically evaluate previous explanations for the persistent eutrophication, and to introduce a new diagnosis and paradigm for the causes and processes behind eutrophication of the Baltic Sea. According to the current consensus, anthropogenic nutrient loading is nearly the sole cause of eutrophication and regular cyanobacterial blooms. However, this study shows that the areal phosphorus loading rate, when modeled properly, is surprisingly low, and unlikely to be the primary cause of eutrophication. Instead, the frequency of the salt water pulses has decreased dramatically during the past 40 years. This is the root cause of eutrophication, via the hyper-vicious cycle of the hypoxic and finally anoxic conditions of the deeps causing internal phosphorus loading, denitrification, and nitrogen and carbon fixation. Furthermore, this work confirms that nitrogen fixation increases in low nitrogen conditions, further increasing eutrophication and cyanobacterial blooms. Thus, the most effective way to break the cycle of eutrophication is to improve the oxygen conditions of the deeps, which really is impossible to achieve by decreasing external loading alone. A key result of this work is that natural processes, rather than human activity, plays a decisive role in the eutrophication process — a perspective that typically faces substantial resistance. This thesis discusses how sociological and political views have affected the scientific community and its pursuit to model the mechanisms of eutrophication of the Baltic Sea. In conclusion, this study leads to important novel insights by providing new models for calculating the external and internal phosphorus loads of the Baltic Sea, with results highlighting the importance of natural processes of internal loading from th, Tiivistelmä Itämeri rehevöityy edelleen, vaikka fosforikuormitusta on vähennetty 80 % tavoitellusta. On siis syytä tutkia miksi Itämeren tila ei ole parantunut. Syntyneen ristiriidan ratkaisemiseksi tarkoituksena on etsiä aiemmista tulkinnoista ristiriitoja, korjata ne ja uudistaa tietopohja uudeksi ja toimivammaksi paradigmaksi. Virhetulkintojen tunnistamiseen sovelletaan Popperin falsifikaatiomenettelyn periaatteita. Konsensuksen mukaan ihmisperäinen kuormitus on lähes yksinomainen syy (sinilevä)rehevyyteen. Kuitenkin Vollenweiderin mallin mukaan tehty, hydrologialla painotettu fosforin pintakuormitus on ollut 1970–1980-luvuilla vain lievää rehevyyttä edellyttävällä tasolla. Ulkoinen kuormitus ei siten voi olla suurin rehevyyden aiheuttaja. Sinilevärehevyyttä ylläpitävänä päätekijänä on syvävesiin happea tuovien suolavesipulssien toistuvuuden romahtaminen. Siitä syntyneet syvävesien ja -pohjien hapettomuudet aiheuttavat Gotlanninmeressä sekä sisäistä kuormitusta että lisärehevöittävää typen- ja hiilensidontaa. Yhdessä näitä prosesseja nimitetään nyt supernoidankehäksi. Johtopäätöksiä: • Ihmistieteelliset ja ympäristöpoliittiset näkemykset vaikuttavat luonnontieteellisiin tulkintoihin luultua enemmän. • Suolavesipulssien harventuminen on hapettomuus- ja rehevyyskierteen perussyy. • Syvänteiden hapettomuus on kaikkialle negatiivisesti säteilevä keskeistekijä. • Itämeren supernoidankehä on purettava saattamalla syvänteet hapellisiksi. • Hapellisuutta ei voida saada aikaan ulkoisen kuormituksen vähentämisellä. • Jäljelle jäävät siten teknologiset keinot, joista toteutuskelvollisimmalta näyttää Itämeren hapellisimman ja kylmimmän veden pumppaus 30 metrin syvyydestä syvänteisiin, mikä tehostaa myös pulssivesien virtausta syvänteiden pohjalle. Tämä väitöskirja sisältää viisi vallitsevasta paradigmasta poikkeavaa tulosta: 1. Itämerelle on kehitetty rehevyyden uusi diagnoosi ja paradigma, 2. Luonto dominoi Itämeren tilaa, ei ihminen, 3. Typensidonnalla on lisärehevöittävä

Emotions modulate ECG signals such that they might affect ECG-based biometric identification in real life application. It motivated in finding good feature extraction methods where the emotional state of the subjects has minimum impacts. This paper evaluates feature extraction based on bivariate empirical mode decomposition (BEMD) for biometric identification when emotion is considered. Using the ECG signal from the Mahnob-HCI database for affect recognition, the features were statistical distributions of dominant frequency after applying BEMD analysis to ECG signals. The achieved accuracy was 99.5% with high consistency using kNN classifier in 10-fold cross validation to identify 26 subjects when the emotional states of the subjects were ignored. When the emotional states of the subject were considered, the proposed method also delivered high accuracy, around 99.4%. We concluded that the proposed method offers emotion-independent features for ECG-based biometric identification. The proposed method needs more evaluation related to testing with other classifier and variation in ECG signals, e.g. normal ECG vs. ECG with arrhythmias, ECG from various ages, and ECG from other affective databases.

We propose new features for emotion recognition from short ECG signals. The features represent the statistical distribution of dominant frequencies, calculated using spectrogram analysis of intrinsic mode function after applying the bivariate empirical mode decomposition to ECG. KNN was used to classify emotions in valence and arousal for a 3-class problem (low-medium-high). Using ECG from the Mahnob-HCI database, the average accuracies for valence and arousal were 55.8% and 59.7% respectively with 10-fold cross validation. The accuracies using features from standard Heart Rate Variability analysis were 42.6% and 47.7% for valence and arousal respectively for the 3-class problem. These features were also tested using subject-independent validation, achieving an accuracy of 59.2% for valence and 58.7% for arousal. The proposed features also showed better performance compared to features based on statistical distribution of instantaneous frequency, calculated using Hilbert transform of intrinsic mode function after applying standard empirical mode decomposition and bivariate empirical mode decomposition to ECG. We conclude that the proposed features offer a promising approach to emotion recognition based on short ECG signals. The proposed features could be potentially used also in applications in which it is important to detect quickly any changes in emotional state.

This research studied violence detection from less than 6-second ECG signals. Features were calculated based on the Bivariate Empirical Mode Decomposition (BEMD) and the Recurrence Quantification Analysis (RQA) applied to ECG signals from violence simulation in a primary school, involving 12 pupils from two grades. The feature sets were fed to a kNN classifier and tested using 10-fold cross validation and leave-one-subject-out (LOSO) validation in subject-dependent and subject-independent training models respectively. Features from BEMD outperformed the ones from RQA in both 10-fold cross validation, i.e. 88% vs. 73% (2nd grade pupils) and 87% vs. 81% (5th grade pupils), and LOSO validation, i.e. 77% vs. 75% (2nd grade pupils) and 80% vs. 76% (5th grade pupils), but have larger variation than the ones from RQA in both validations. Average performances for subject-specific system in 10-fold cross validation were 100% vs. 93% (2nd grade pupils) and 100% vs. 97% (5th grade pupils) for features from the BEMD and the RQA respectively. The results indicate that ECG signals as short as 6 seconds can be used successfully to detect violent events using subject-specific classifiers.

Dimensionality reduction (DR) is an important issue in classification and pattern recognition process. Using features with lower dimensionality helps the machine learning algorithms work more efficient. Besides, it also can improve the performance of the system. This paper explores supervised dimensionality reduction, LDA (Linear Discriminant Analysis), NCA (Neighbourhood Components Analysis), and MCML (Maximally Collapsing Metric Learning), in emotion recognition based on ECG signals from the Mahnob-HCI database. It is a 3-class problem of valence and arousal. Features for kNN (k-nearest neighbour) are based on statistical distribution of dominant frequencies after applying a bivariate empirical mode decomposition. The results were validated using 10-fold cross and LOSO (leave-one-subject-out) validations. Among LDA, NCA, and MCML, the NCA outperformed the other methods. The experiments showed that the accuracy for valence was improved from 55.8% to 64.1%, and for arousal from 59.7% to 66.1% using 10-fold cross validation after transforming the features with projection matrices from NCA. For LOSO validation, there is no significant improvement for valence while the improvement for arousal is significant, i.e. from 58.7% to 69.6%.

Multimodal measurements are increasingly being employed in the study of human physiology. Brain studies in particular can draw advantage of simultaneous measurements using different modalities to analyse correlations, mechanisms and relationships of physiological signals and their dynamics in relation to brain functions. Moreover, multimodal measurements help to identify components of physiological dynamics generated specifically by the brain. This thesis summarizes the study, design and development of non-invasive medical instruments that can be utilized in conjunction with magnetic resonance imaging (MRI). A key challenge in the development of measurement methods is posed by the extraordinary requirements that the MRI environment poses - all materials need to be MR-compatible and the selected instruments and devices must not be affected by the strong magnetic field generated by the MRI scanner nor the MRI by the instruments placed within its scanning volume. The presented methods allow simultaneous continuous measurement of heart rate (HR) and metabolism from the brain cortex as well as pulse wave velocity (PWV) and blood pressure measurements in synchrony with electroencephalography (EEG) and MRI. Furthermore, the thesis explored the reliability and accuracy of the responses gathered by the developed instruments and, using new experimental methods, estimated the propagation of near-infrared light in the human brain. The goal of the novel multimodal measurement environment is to provide more extensive tools for medical researchers, neurologists in particular, to acquire accurate information on the function of the brain and the human body. Measurements have been performed on more than 70 persons using the presented multimodal setup to study such factors as the correlation between blood oxygen level-dependent (BOLD) data and low-frequency oscillations (LFOs) during the resting state., Tiivistelmä Multimodaalisia kuvantamismenetelmiä käytetään enenevässä määrin ihmisen fysiologian ja elintoimintojen tutkimisessa. Erityisesti aivotutkimuksessa samanaikaisesti useammalla modaliteetilla mittaaminen mahdollistaa erilaisten fysiologisten mekanismien ja niiden korrelaatioiden tutkimisen kehon ja aivotoimintojen välillä. Lisäksi multimodaaliset mittaukset auttavat yksilöimään fysiologiset komponentit toisistaan ja identifioimaan aivojen tuottamia fysiologisia signaaleja. Tämä väitöskirja kokoaa tutkimustyön sekä laite- ja instrumentointisuunnittelun ja sen kehittämistyön ei-invasiivisesti toteutettujen lääketieteen mittausmenetelmien käyttämiseksi magneettikuvauksen aikana. Erityishaasteena työssä on ollut magneettikuvausympäristö, joka asettaa erityisvaatimuksia mm. mittalaitteissa käytettäville materiaaleille sekä laitteiden häiriönsiedolle magneettikuvauslaitteen aiheuttaman voimakkaan magneettikentän takia. Kehitettävät mittausmenetelmät eivät myöskään saa aiheuttaa häiriöitä magneettikuvauslaitteen tuottamalle kuvainformaatiolle. Väitöskirjassa esitettävät mittausmenetelmät tekevät mahdolliseksi mitata magneettikuvausympäristössä ihmisen sydämen sykettä, veren virtauksen kulkunopeutta ja verenpaineen vaihteluja sekä aivokuoren metaboliaa — kaikki synkronissa aivosähkökäyrän mittaamisen ja magneettikuvantamisen kanssa. Lisäksi väitöskirjassa tutkitaan kehitettyjen mittausmenetelmien antamaa mittaustarkkuutta sekä arvioidaan lähi-infrapunavalon etenemistä ihmisen aivoissa uudenlaisella menetelmällä. Kehitetyllä multimodaalisella mittausympäristöllä on tavoitteena antaa lääketieteen alan tutkijoille, erityisesti neurologeille, käyttöön mittausmenetelmiä, joiden avulla voidaan tutkia ihmisen aivojen ja kehon välisiä toimintoja aiempaa kattavammin. Laitekokonaisuudella on tutkittu jo yli 70:tä henkilöä. Näissä mittauksissa on tutkittu mm. veren happitasojen hitaita vaihteluja ihmisen aivojen ollessa lepotilassa, ns. resting state -tilassa.

The main objective of this thesis project is to implement a wireless vital signs monitoring system for measuring the ECG of a patient in the home environment. The research focuses on two specific research objectives: 1) the development of a distributed healthcare system for vital signs monitoring using wireless sensor network devices and 2) a practical test and performance evaluation for the reliability for such low-rate wireless technology in ubiquitous health monitoring applications. The first section of the thesis describes the design and implementation of a ubiquitous healthcare system constructed from tiny components for the home healthcare of elderly persons. The system comprises a smart shirt with ECG electrodes and acceleration sensors, a wireless sensor network node, a base station and a server computer for the continuous monitoring of ECG signals. ECG data is a commonly used vital sign in clinical and trauma care. The ECG data is displayed on a graphical user interface (GUI) by transferring it to a PDA or a terminal PC. The smart shirt is a wearable T-shirt designed to collect ECG and acceleration signals from the human body in the course of daily life. In the second section, a performance evaluation of the reliability of IEEE 802.15.4 low-rate wireless ubiquitous health monitoring is presented. Three scenarios of performance studies are applied through practical tests: 1) the effects of the distance between sensor nodes and base-station, 2) the deployment of the number of sensor nodes in a network and 3) data transmission using different time intervals. These factors were measured to analyse the reliability of the developed technology in low-rate wireless ubiquitous health monitoring applications. The results showed how the relationship between the bit-error-rate (BER) and signal-to-noise ratio (SNR) was affected when varying the distance between sensor node and base-station, through the deployment of the number of sensor nodes in a network and t

Analyzing the human body by the application of alternating electrical currents is not a widely known method in medicine. In our research, we stimulated test persons exposing them to different low frequencies and measured their responses to them. This method, known as FAM (Frequency Analysis Method), can be used to estimate the physiological condition of patients. In this study, we present a method of processing the results using neural networks. By producing user-friendly visual data, the processing method aids a physiotherapist in the interpretation of the results, resulting in a more reliable diagnosis., Presented at Annual International Conference of the IEEE Engineering in Medicine and Biology Society (23rd) held in Istanbul, Turkey on 25-28 Oct 2001. See also ADM001351 for entire conference on cd-rom. , The original document contains color images.