Dans cette thèse, on introduit la métrique "Coefficient de forme" pour la classement des données de dissimilitudes. Cette approche est inspirée par l'analyse discriminante géométrique et on a défini des règles de décision pour imiter le comportement du classifieur linéaire et quadratique. Le nombre de paramètres est limité (deux par classe). On a également étendu et amélioré cette démarche avantageuse et rapide pour apprendre uniquement à partir des représentations de dissimilitudes en utilisant l'efficacité du classificateur des Machines à Vecteurs de Support. Comme contexte applicatif pour la classification par dissimilitudes, on utilise la recherche d'images à l'aide d'une représentation des images en multi échelle en utilisant la "Pyramide Réduite Différentielle". Une application pour la description de visages est développée. Des résultats de classification à partir du coefficient de forme et utilisant une version adaptée des Machines à Vecteurs de Support, sur des bases de données issues des applications du monde réel sont présentés et comparés avec d'autres méthodes de classement basées sur des dissimilitudes. Il en ressort une forte robustesse de la méthode proposée avec des perfommances supérieures ou égales aux algorithmes de l'état de l'art. / The dissimilarity representation is an alternative for the use of features in the recognition of real world objects like images, spectra and time-signal. Instead of an absolute characterization of objects by a set of features, the expert or the system is asked to define a measure that estimates the dissimilarity between pairs of objects. Such a measure may also be defined for structural representations such as strings and graphs. The dissimilarity representation is potentially able to bridge structural and statistical pattern recognition. In this thesis we introduce a new fast Mahalanobis-like metric the “Shape Coefficient” for classification of dissimilarity data. Our approach is inspired by the Geometrical Discriminant Analysis and we have defined decision rules to mimic the behavior of the linear and quadratic classifier. The number of parameters is limited (two per class). We also expand and ameliorate this advantageous and rapid adaptive approach to learn only from dissimilarity representations by using the effectiveness of the Support Vector Machines classifier for real-world classification tasks. Several methods for incorporating dissimilarity representations are presented, investigated and compared to the “Shape Coefficient” in this thesis: • Pekalska and Duin prototype dissimilarity based classifiers; • Haasdonk's kernel based SVM classifier; • KNN classifier. Numerical experiments on artificial and real data show interesting behavior compared to Support Vector Machines and to KNN classifier: (a) lower or equivalent error rate, (b) equivalent CPU time, (c) more robustness with sparse dissimilarity data. The experimental results on real world dissimilarity databases show that the “Shape Coefficient” can be an alternative approach to these known methods and can be as effective as them in terms of accuracy for classification.

Organizations in Jamaica have been impacted by globalization and the opportunities and challenges of cultural incompatibilities. Most previous studies on cultural incompatibilities have focused on the impact on expatriates leaving a gap in the literature with respect to the implications for host country nationals, and specifically Jamaicans. This quantitative study focused on employees of 2 companies in Jamaica, an energy company and a hospitality company. It examined cultural dissimilarity with respect to host country nationals and expatriates, and its effect on the productivity, job satisfaction, affective commitment, and normative commitment of these employees (N = 110). In addition to the above variables, the study also identified the role that gender, age, and tenure played in these relationships. Diversity theory, social exchange theory, homophily, and repulsion hypothesis formed the theoretical framework for this study, and multiple regression and correlation were utilized in the analysis of the data collected. The results of the study indicated correlation and predictive relationships between/among: culture and job satisfaction; age, gender, and experience in relation to job satisfaction; age, gender, and experience in relation to affective commitment; and culture, age, gender, and experience in relation to affective commitment. Social change implications for this study include the development of country-specific culture awareness training programs for both host country nationals and expatriates. It is further expected that the findings of this study will increase knowledge on the subject and help in the development of human resource management policies and procedures. These policies should aid in improved job attitudes and productivity for host country nationals.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第18482号 / 情博第533号 / 新制||情||94(附属図書館) / 31360 / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 美濃 導彦, 教授 河原 達也, 教授 中村 裕一 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

No description available.

In this manuscript I examine outcomes associated with social networks in organizations. Specifically, I consider how two characteristics of social networks (i.e., centrality, tie strength) can affect the performance and satisfaction of employees at work. Then, I explore the role that perceptions of fit (i.e., person-group fit, person-organization) may play in mediating the relationship between social network characteristics and (a) employee performance and (b) job satisfaction. Moreover, I investigate boundary conditions of the aforementioned mediated relationships (i.e., social network characteristics - fit perceptions - employee performance; social network characteristics - fit perceptions - job satisfaction). First, I consider how individual differences (i.e., racioethnicity, sex) generate employee dissimilarity that likely moderates the relationship between structural network characteristics and perceived fit in the mediated relationships proposed. Second, I examine an organizational variable (i.e., perceived diversity climate) as a first and second stage moderator of the aforementioned mediated relationships. Overall, it is necessary to investigate the relationships proposed in the model, because studying social networks helps us to understand why employees interact with certain individuals (or not with others) and how organizational outcomes are affected by employees' choices regarding their social networks. / Business Administration/Human Resource Management

Det som ofta faller bort i den offentliga debatten är att segregationens innebörd anspelar på åtskillnader av olika grupper, inte enbart de ”utsatta” utan även de socioekonomiskt starkare grupperna. Denna studie kommer att undersöka hur segregationen ser ut inom Sundsvalls tätortsområden och vilka faktorerna är som har störst påverkan till de skillnader som finns. Studien baseras på att urskilja den relativa segregationen, d.v.s. fördelningen av segregationspåverkande faktorer, i positiv och negativ riktning. Syftet i studien är sedan att jämföra två olika geografiska indelningssystem. Det ena är det kommunala nyckelkodssystemet (NYKO) och det andra är Statistiska centralbyråns regionala indelningssystem, demografiska statistikområden (DeSO). I metoden har två olika index använts för att mäta fördelningen av segregationsfaktorer genom en multikriterieanalys (MKA). Den första mätningen har skett genom en nyutvecklad segregationsindex och den andra mätningen genom index of dissimilarity. Studien har visat att det finns svårigheter kring att använda ett indelningssystem som kan verka funktionellt i alla avseenden. Beroende på vad studien syftar till att mäta så spelar olika zon- och skalindelningar en stor roll i hur resultatet framställs. Resultatet visar att segregationen utspelar sig inom både de socioekonomiskt svaga och starka områdena. Det finns däremot svårigheter med att bedöma vilken områdesindelning som är mest användbar då de verkar på olika grunder. Genom att jämföra DeSO och NYKO har resultaten av studien visat att befolkningsantalet och storleken på den geografiska områdesindelningen har en stor betydelse för hur pålitlig en studie kan bli. / What often falls away in the public debate is that the meaning of the segregation alludes to the separation of different groups, not just the "vulnerable" but also the socio-economically stronger groups. This study will investigate how the segregation plays out within Sundsvall's urban areas and which factors have the greatest impact on the differences that exist. The study is based on distinguishing the relative segregation, i. e. the distribution of factors affecting segregation, in a positive and negative direction. The purpose of the study is then to compare two different geographical area systems. One is the municipal key code system (NYKO) and the other is the state regional area system, demographic statistics areas (DeSO). In the method, two different indexes have been used to measure the distribution of segregation factors through a multi-criteria decision analysis (MKA). The first measurement has been made by a newly developed segregation index and the second measurement by the index of dissimilarity. The study has shown that there are difficulties in finding an area system that can function efficiently for all purposes. Depending on what the study aims to measure, the different zone and scale configurations play a major role in how the result is produced. The result of the study shows that the segregation takes place in both the socio-economically weak and strong areas. There are, on the other hand, difficulties in assessing which of the two area systems that is the most practical since they both operate on different grounds. By comparing DeSO and NYKO, the results of the study have shown that the population and size of the geographical area unit are of great importance for how reliable a study can be.

Our vision is as yet unsurpassed by machines because of the sophisticated representations of objects in our brains. This representation is vastly different from a pixel-based representation used in machine storages. It is this sophisticated representation that enables us to perceive two faces as very different, i.e, they are far apart in the “perceptual space”, even though they are close to each other in their pixel-based representations. Neuroscientists have proposed distances between responses of neurons to the images (as measured in macaque monkeys) as a quantification of the “perceptual distance” between the images. Let us call these neuronal dissimilarity indices of perceptual distances. They have also proposed behavioural experiments to quantify these perceptual distances. Human subjects are asked to identify, as quickly as possible, an oddball image embedded among multiple distractor images. The reciprocal of the search times for identifying the oddball is taken as a measure of perceptual distance between the oddball and the distractor. Let us call such estimates as behavioural dissimilarity indices. In this thesis, we describe a decision-theoretic model for visual search that suggests a connection between these two notions of perceptual distances. In the first part of the thesis, we model visual search as an active sequential hypothesis testing problem. Our analysis suggests an appropriate neuronal dissimilarity index which correlates strongly with the reciprocal of search times. We also consider a number of alternative possibilities such as relative entropy (Kullback-Leibler divergence), the Chernoff entropy and the L1-distance associated with the neuronal firing rate profiles. We then come up with a means to rank the various neuronal dissimilarity indices based on how well they explain the behavioural observations. Our proposed dissimilarity index does better than the other three, followed by relative entropy, then Chernoff entropy and then L1 distance. In the second part of the thesis, we consider a scenario where the subject has to find an oddball image, but without any prior knowledge of the oddball and distractor images. Equivalently, in the neuronal space, the task for the decision maker is to find the image that elicits firing rates different from the others. Here, the decision maker has to “learn” the underlying statistics and then make a decision on the oddball. We model this scenario as one of detecting an odd Poisson point process having a rate different from the common rate of the others. The revised model suggests a new neuronal dissimilarity index. The new dissimilarity index is also strongly correlated with the behavioural data. However, the new dissimilarity index performs worse than the dissimilarity index proposed in the first part on existing behavioural data. The degradation in performance may be attributed to the experimental setup used for the current behavioural tasks, where search tasks associated with a given image pair were sequenced one after another, thereby possibly cueing the subject about the upcoming image pair, and thus violating the assumption of this part on the lack of prior knowledge of the image pairs to the decision maker. In conclusion, the thesis provides a framework for connecting the perceptual distances in the neuronal and the behavioural spaces. Our framework can possibly be used to analyze the connection between the neuronal space and the behavioural space for various other behavioural tasks.

Les travaux de cette thèse ont été initiés par des problèmes d’apprentissage de données radiomiques. La Radiomique est une discipline médicale qui vise l’analyse à grande échelle de données issues d’imageries médicales traditionnelles, pour aider au diagnostic et au traitement des cancers. L’hypothèse principale de cette discipline est qu’en extrayant une grande quantité d’informations des images, on peut caractériser de bien meilleure façon que l’œil humain les spécificités de cette pathologie. Pour y parvenir, les données radiomiques sont généralement constituées de plusieurs types d’images et/ou de plusieurs types de caractéristiques (images, cliniques, génomiques). Cette thèse aborde ce problème sous l’angle de l’apprentissage automatique et a pour objectif de proposer une solution générique, adaptée à tous problèmes d’apprentissage du même type. Nous identifions ainsi en Radiomique deux problématiques d’apprentissage: (i) l’apprentissage de données en grande dimension et avec peu d’instances (high dimension, low sample size, a.k.a.HDLSS) et (ii) l’apprentissage multi-vues. Les solutions proposées dans ce manuscrit exploitent des représentations de dissimilarités obtenues à l’aide des Forêts Aléatoires. L’utilisation d’une représentation par dissimilarité permet de contourner les difficultés inhérentes à l’apprentissage en grande dimension et facilite l’analyse conjointe des descriptions multiples (les vues). Les contributions de cette thèse portent sur l’utilisation de la mesure de dissimilarité embarquée dans les méthodes de Forêts Aléatoires pour l’apprentissage multi-vue de données HDLSS. En particulier, nous présentons trois résultats: (i) la démonstration et l’analyse de l’efficacité de cette mesure pour l’apprentissage multi-vue de données HDLSS; (ii) une nouvelle méthode pour mesurer les dissimilarités à partir de Forêts Aléatoires, plus adaptée à ce type de problème d’apprentissage; et (iii) une nouvelle façon d’exploiter l’hétérogénèité des vues, à l’aide d’un mécanisme de combinaison dynamique. Ces résultats ont été obtenus sur des données radiomiques mais aussi sur des problèmes multi-vue classiques. / The work of this thesis was initiated by a Radiomic learning problem. Radiomics is a medical discipline that aims at the large-scale analysis of data from traditional medical imaging to assist in the diagnosis and treatment of cancer. The main hypothesis of this discipline is that by extracting a large amount of information from the images, we can characterize the specificities of this pathology in a much better way than the human eye. To achieve this, Radiomics data are generally based on several types of images and/or several types of features (from images, clinical, genomic). This thesis approaches this problem from the perspective of Machine Learning (ML) and aims to propose a generic solution, adapted to any similar learning problem. To do this, we identify two types of ML problems behind Radiomics: (i) learning from high dimension, low sample size (HDLSS) and (ii) multiview learning. The solutions proposed in this manuscript exploit dissimilarity representations obtained using the Random Forest method. The use of dissimilarity representations makes it possible to overcome the well-known difficulties of learning high dimensional data, and to facilitate the joint analysis of the multiple descriptions, i.e. the views.The contributions of this thesis focus on the use of the dissimilarity easurement embedded in the Random Forest method for HDLSS multi-view learning. In particular, we present three main results: (i) the demonstration and analysis of the effectiveness of this measure for HDLSS multi-view learning; (ii) a new method for measuring dissimilarities from Random Forests, better adapted to this type of learning problem; and (iii) a new way to exploit the heterogeneity of views, using a dynamic combination mechanism. These results have been obtained on radiomic data but also on classical multi-view learning problems.

The city of Richmond has practiced mixed land-use policies to encourage non-private-vehicle commuting for decades based on the successful examples or the empirical evidence of other cities. However, the idea violates one of common logical fallacy—“all things are equal.” Using the indices of land-use diversity, this study explores the correlation between land-use mixture and home-based trip for the city of Richmond. This paper calculates two common indices of land-use mixture—entropy, and dissimilarity. The results indicate that although Richmond’s land-use mixture and home-based trip do have a correlation, the correlation is weak. One possible reason is that socioeconomic actors have a stronger influence on transportation than land-use mixture. However, this assumption still needs further analysis in order to be verified.

Uma grande parte das florestas secundárias ao redor do mundo é resultado da regeneração natural de áreas agrícolas abandonadas, localizadas em paisagens altamente modificadas pelo homem. O conhecimento sobre a composição funcional dessas florestas ainda é escasso, sendo urgentes pesquisas que contribuem para esse entendimento, pois os atributos funcionais das plantas são condutores da dinâmica florestal e, portanto, importantes para a manutenção da biodiversidade e dos serviços ecossistêmicos. Nesse contexto, escolhemos como modelo uma bacia hidrográfica no sudeste do Brasil, com matriz agrícola e florestas sob diferentes condições ambientais, para responder a duas perguntas: 1) Como fatores temporais, locais e de paisagem afetam a composição funcional de florestas secundárias (regenerando sob plantios de eucalipto e pastos abandonados) em paisagens agrícolas e 2) Como se comportam as diversidades taxonômica e funcional em tipos florestais com diferentes históricos de vida (florestas remanescentes conservadas, florestas remanescentes degradadas, florestas secundárias regenerando sob plantios de eucalipto abandonados e florestas secundárias regenerando sob pastos abandonados). Para responder as duas perguntas utilizamos o valor médio por espécie de dados primários (área foliar, área foliar específica, conteúdo de matéria seca da folha e espessura da folha) e secundários (densidade da madeira). A partir disso, usamos modelos generalizados mistos, média ponderada pela comunidade para cada atributo funcional e três índices de diversidade funcional para responder a primeira pergunta e testes de dissimilaridade para responder a segunda. Ao todo foram avaliadas 59 parcelas, 43 de florestas secundárias, 10 de florestas remanescentes degradadas e seis de florestas remanescentes conservadas, totalizando 6.089 indivíduos levantados e 284 espécies identificadas. Concluímos que 1) área foliar, área foliar específica e riqueza funcional de florestas secundárias são afetadas, ao mesmo tempo e de formas diferentes, por idade, declividade, teor de argila do solo, área basal de eucaliptos, cobertura florestal média, diferença na cobertura florestal e proximidade com cana- de-açúcar; e 2) florestas secundárias tendem a ter maior riqueza funcional e taxonômica que florestas conservadas, sendo que dentro dos tipos florestais existe uma diversidade beta taxonômica maior que a funcional, com comunidades apresentando, em geral, espécies abundantes diferentes com atributos funcionais similares. Nossos resultados demonstram que as florestas secundárias são afetadas tanto por fatores naturais quanto por fatores antrópicos, os quais devem ser levados em consideração tanto em pesquisas que visam compreender os condutores da regeneração natural quanto em projetos que visam utilizar esse processo como estratégia para conservar a biodiversidade e prover serviços ecossistêmicos. Adicionalmente, demonstramos que florestas remanescentes degradadas e secundárias são importantes fontes de biodiversidade e, portanto, potenciais provedoras de serviços ecossistêmicos / Large part of the secondary forests across the world is the result of the natural regeneration in abandoned agricultural areas located in landscapes highly human-modified. Knowledge about the functional composition of these forests is still scarce, and research contributing to this understanding results urgent since the functional traits may be important drivers of the forest ecological dynamics and therefore are important for the maintenance of biodiversity and ecosystem services the forest provides. In this context, we chose as a study case a watershed in the Atlantic Forest in Southeastern Brazil, with an agricultural matrix and patches of secondary forests regenerating under different environmental conditions, to answer the following two questions: 1) How temporal, local and landscape factors affect the functional composition of secondary forests regenerating in Eucalyptus plantations and abandoned pastures and (2) how taxonomic and functional diversity behave in forest types under different land-use histories (i.e. conserved remnant forest, degraded remnant forest, secondary forests regenerating in abandoned Eucalyptus plantations and secondary forests regenerating in abandoned pastures). In order to answer the two questions we used the mean value per species of primary data (leaf area, specific leaf area, leaf dry matter content and leaf thickness) and secondary data (wood density). We use mixed generalized models, the community weighted mean with species mean traits and three diversity indexes to answer the first question and dissimilarity tests to answer the second question. Overall, 59 plots were evaluated, including 43 secondary forests, 10 degraded remnant forests and 6 conserved remnant forest, where we registered totaling total of 6.089 individuals and 284 species. We conclude that: 1) leaf area, specific leaf area and functional richness of secondary forests is affected, at the same time and in different ways, by age, slope, soil clay content, basal area of Eucalyptus, average native forest cover , difference in surrounding native forest cover and proximity to sugarcane plantations; and 2) secondary forests tend to have higher functional and taxonomic richness than conserved forests, and within the forest types taxonomic beta diversity results higher than functional, with communities presenting, in general, different abundant species with similar functional traits. Our results demonstrate that secondary forests are affected by both natural and anthropogenic factors, which should be taken into account both in research aimed at understanding the drivers of natural regeneration and in projects that use this process as a strategy to conserve biodiversity and provide ecosystem services. In addition, we show that degraded remnants and secondary forests are important sources of biodiversity and therefore potential providers of ecosystem services.