Cette thèse, s’inscrivant dans une approche sociocognitive, étudie l’apprentissage professionnel dans une formation d’enseignants du 2nd degré se spécialisant pour scolariser les élèves en situation de handicap. Son objectif est de montrer que l’enseignant en formation de spécialisation construit des savoirs professionnels dans l’interaction sociale en présentiel et en ligne avec ses pairs par le biais d’outils du dispositif hybride, en particulier à travers un carnet de bord en ligne. Notre cadre théorique mobilise les théories de l’apprentissage professionnel. La construction sociale des apprentissages prend en compte l’importance de l’environnement social, culturel et la place de l’individu dans la construction des savoirs, tout en mobilisant les interactions via différentes modalités. Les savoirs professionnels peuvent être mobilisés par l’enseignant et repérés en tant que savoirs énoncés dans le carnet de bord en ligne, en tant que savoirs perçus par l’évolution du niveau du sentiment d’efficacité professionnelle (SEPro) et en tant que savoirs constatés dans les pratiques enseignantes d’adaptation en classe et hors la classe. À partir d’une méthodologie mixte, cette recherche présente les résultats d’une enquête menée auprès de cinq enseignants du 2nd degré poursuivant une formation de spécialisation (2CA-SH) dans un contexte hybride à l’ESPE de Toulouse. Une comparaison est conduite en parallèle sur l’évolution du niveau de SEPro avec un groupe d’enseignants suivant la même formation en présentiel (n= 24) et un groupe sans formation (n =58). L’analyse des éléments empiriques montre que les interactions en ligne et présentielles entre pairs permettent de construire des savoirs professionnels relatifs aux gestes d’adaptation et à la connaissance des élèves et que ces savoirs se diversifient sur l’année. Les savoirs construits sont mobilisés dans leur SEPro, leurs pratiques et dans l’élaboration du carnet en ligne. La mise en relation de la nature des savoirs professionnels et des processus de construction à l’œuvre (apprentissage social) permet de mettre au jour les processus privilégiés par les enseignants interagissant entre pairs en présentiel et en ligne et de proposer un modèle de l’apprentissage social en contexte de formation hybride. / This thesis, in a social cognitive approach, studies the professional learning of secondary education teachers specializing in disabled pupils. The goal is to show that under specializing learning teachers build professional knowledge in the social interaction whether on-site or on-line with their peers thanks to the hybrid system tools; especially thanks to an on-line log book. Our theoretical framework gathers professional learning theories. The social construction of learning takes into account the importance of the social and cultural environments, the individual’s place when building knowledge, as well as the interactions through different aspects. The professional knowledge can be used by the teacher and be seen as knowledge mentioned in the on-line log book, as the perceived knowledge through the evolution of the Professional Feeling of Efficiency level (SEPro in French), as well as the perceived knowledge in the teacher’s practice when adapting to class and outside the classroom. From a mixed methodology, this study presents the survey results performed following five teachers from the secondary education under a specializing learning (2CA-SH: Accreditation for specialized teachers for secondary school and higher levels) in a hybrid context at the ESPE of Toulouse (Teaching and Learning Graduate School). A comparison is done parallel to the SEPro level evolution with a group of teachers following the same on-site training (n= 24) and a group without training (n =58). The analysis of the empirical elements shows that the on-line and on-site interactions with peers let them build professional knowledge linked to adaptive professional behaviors and the pupils understanding. It also shows that knowledge becomes diversified during the learning year. The built knowledge is used in their SEPro, in their practice and in the elaboration of the on-line log book. The link between the professional knowledge nature and the learning process (social learning) allows to highlight the processes the teachers interacting on-site and on-line with peers prefer. It also allows to propose a social learning model in the context of a hybrid learning. / Esta tesis, que se inscribe en una lógica socio cognitiva, estudia el aprendizaje profesional en el marco de una formación de profesores del grado Superior especializados en el tratamiento de alumnos con discapacidad. Su objetivo consiste en mostrar que el profesor en formación especializada construye conocimientos profesionales en la interacción social en presencial y a distancia con sus colegas gracias a herramientas del dispositivo híbrido, en especial a través de un diario de a bordo en línea. Nuestro marco teórico moviliza las teorías del aprendizaje profesional. La construcción social de los aprendizajes toma en cuenta la importancia del entorno social, cultural y el lugar del individuo en la construcción de los conocimientos, al mismo tiempo que moviliza las interacciones a través de diferentes modalidades. Los conocimientos profesionales pueden ser movilizados por el profesor e identificados como conocimientos listados en el diario de a bordo en línea, como conocimientos percibidos por la evolución del nivel de sentimiento de eficiencia profesional (SEPro) y como conocimientos constatados en las prácticas de adaptación en clase y fuera de clases. A partir de una metodología mixta, esta investigación presenta los resultados de una encuesta realizada con cinco profesores del grado Superior que siguieron una formación especializada (2CA-SH: Certificado de especialización complementaria del grado Superior) en un contexto híbrido en la ESPE (Escuela Superior de Profesorado y de la Educación) de Tolosa. Un comparativo se realizó en paralelo con la evolución del nivel de SEPro con un grupo de profesores que realizaron la misma formación presencial (n= 24) y un grupo sin formación (n =58). El análisis de los elementos empíricos muestra que las interacciones en línea y presenciales entre colegas permiten construir conocimientos profesionales relativos a los gestos de adaptación y al conocimiento de los alumnos y que dichos conocimientos se diversifican conforme pasa el año. Los conocimientos construidos son movilizados en su SEPro, en sus prácticas y en la elaboración del diario de a bordo. La interconexión de la naturaleza de los conocimientos profesionales y de los procesos de construcción de obras (aprendizaje social) permite actualizar los procesos privilegiados por los profesores que interactúan entre ellos en presencial y en línea y proponer un modelo de aprendizaje social en contexto de formación híbrida.

La complexité inhérente à la coordination des mouvements d'un groupe de robots mobiles est traitée en investiguant plus avant les potentialités des architectures de contrôle comportementales dont le but est de briser la complexité des tâches à exécuter. En effet, les robots mobiles peuvent évoluer dans des environnements très complexes et nécessite de surcroît une coopération précise et sécurisée des véhicules pouvant rapidement devenir inextricable. Ainsi, pour maîtriser cette complexité, le contrôleur dédié à la réalisation de la tâche est décomposé en un ensemble de comportements/contrôleurs élémentaires (évitement d'obstacles et de collision entre les robots, attraction vers une cible, etc.) qui lient les informations capteurs (provenant de caméras, des capteurs locaux du robot, etc.) aux actionneurs des différentes entités robotiques. La tâche considérée est la navigation en formation en présence d'obstacles (statiques et dynamiques). La spécificité de l'approche théorique consiste à allier les avantages des architectures de contrôle comportementales à la méthode de la structure virtuelle où le groupe de robots mobiles suit un corps virtuel avec une dynamique (vitesse, direction) donnée. Ainsi, l'activation d'un comportement élémentaire en faveur d'un autre se fait en respectant les contraintes structurelles des robots (e.g. vitesses et accélérations maximales, etc.) en vue d'assurer le maximum de précision et de sécurité des mouvements coordonnés entre les différentes entités mobiles. La coopération consiste à se partager les places dans la structure virtuelle de manière distribuée et de façon à atteindre plus rapidement la formation désirée. Pour garantir les critères de performances visés par l'architecture de contrôle, les systèmes hybrides qui permettent de commander des systèmes continus en présence d'évènements discrets sont exploités. En effet, ces contrôleurs (partie discrète) permettent de coordonner l'activité des différents comportements (partie continue) disponibles au niveau de l'architecture, tout en offrant une analyse automaticienne rigoureuse de la stabilité de celle-ci au sens de Lyapunov. Chaque contribution est illustrée par des résultats de simulation. Le dernier chapitre est dédié à l'implémentation de l'architecture de contrôle proposée sur un groupe de robots mobiles Khepera III. / Inherent difficulty of coordinating a group of mobile robots is treated by investigating behavior-based architectures which aim to break task complexity. In fact, multi-robot navigation may become rapidly inextricable, specifically if it is made in hazardous and dynamical environment. The considered task is the navigation in formation in presence of (static and dynamic) obstacles. To overcome its complexity, it is proposed to divide the overall task into two basic behaviors/controllers (obstacle avoidance, attraction to a dynamical target). Applied control is chosen among these controllers according to sensors information (camera, local sensors, etc.). Theoretic approach combines behavior-based and the virtual structure strategy which considers the formation as a virtual body with a given dynamic (velocity, direction). Thus, activating a controller or another is accomplished while respecting structural robots constraints (e.g. maximal velocities and accelerations). The objective is to insure the highest precision and safety of the coordinated motion between the robots. These ones cooperate by optimizing the way of sharing their places in the formation in order to form it in a faster manner. To guarantee performance criteria of the control architecture, hybrid systems tolerating the control of continuous systems in presence of discrete events are explored. In fact, this control allows coordinating (by discrete part) the different behaviors (continuous part) of the architecture. A complete analysis of this architecture stability is also given thanks to Lyapunov-based theory. Every contribution is illustrated through simulation results. The last chapter is devoted to the implementation of the proposed control architecture on a group of Khepera III robots.

O presente trabalho apresenta dois novas modelos conexionistas, baseados na teoria da adaptação ressonante (ART): Simplified Fuzzy ARTMAP e Semantic ART (SMART). Descreve-se a modelagem, adaptação, implementação e validação destes, enquanto incorporados ao sistema hibrido HYCONES, para resolução de problemas de diagnostico medico em cardiopatias congênitas e nefrologia. HYCONES é uma ferramenta para a construção de sistemas especialistas híbridos que integra redes neurais com frames, assimilando as qualidades inerentes aos dois paradigmas. 0 mecanismo de frames fornece tipos construtores flexíveis para a modelagem do conhecimento do domínio, enquanto as redes neurais, representadas na versão original de HYCONES pelo modelo neural combinatório (MNC), possibilitam tanto a automação da aquisição de conhecimento, a partir de uma base de casos, quanta a implementação de aprendizado indutivo e dedutivo. A teoria da adaptação ressonante 6 caracterizada, principalmente, pela manutenção do equilíbrio entre as propriedades de plasticidade e estabilidade durante o processo de aprendizagem. ART inclui vários modelos conexionistas, tais como: Fuzzy ARTMAP, Fuzzy ART, ART 1, ART 2 e ART 3. Dentre estes, a rede neural Fuzzy ARTMAP destaca-se por possibilitar o tratamento de padr6es analógicos a partir de dois módulos ART básicos. O modelo Simplified Fuzzy ARTMAP, como o pr6prio nome o diz, a uma simplificação da rede neural Fuzzy ARTMAP. Ao contrario desta, o novo modelo possibilita o tratamento de padrões analógicos, a partir de apenas um modulo ART, responsável pelo tratamento dos padrões de entrada, adicionado de uma camada, responsável pelos padrões alvo. Mesmo com apenas um modulo ART, o modelo Simplified Fuzzy ARTMAP 6 capaz de reter o mesmo nível de desempenho obtido com a rede neural Fuzzy ARTMAP pois, continua a garantir, conjuntamente, a maximização da generalização e a minimização do erro preditivo, através da execução da estratégia match-tracking. Para a construção da base de casos de cardiopatias congênitas, 66 prontuários médicos, das três cardiopatias congênitas mais freqüentes, foram extraídos do banco de dados de pacientes submetidos a cirurgia cardíaca no Instituto de Cardiologia RS (ICFUC-RS). Tais prontuários abrangem o período de janeiro de 1986 a dezembro de 1990 e reportam 22 casos de Comunicação Interatrial (CIA), 29 de Comunicação Interventricular (CIV) e 15 de Defeito Septal Atrioventricular (DSAV). Para a análise de desempenho do sistema, 33 casos adicionais, do referido período, foram extraídos aleatoriamente do banco de dados do ICFUC-RS. Destes 33 casos, 13 apresentam CIA, 10 CIV e 10 DSAV. Para a construção da base de casos de síndromes renais, 381 prontuários do banco de dados de síndromes renais da Escola Paulista de Medicina foram analisados e 58 evidencias, correspondentes a dados de hist6ria clinica e exame físico dos pacientes, foram extraídas semi-automaticamente. Do total de casos selecionados, 136 apresentam Uremia, 85 Nefrite, 100 Hipertensão e 60 Litiase. Dos 381 casos analisados, 254 foram escolhidos aleatoriamente para a composicao do conjunto de treinamento, enquanto que os demais foram utilizados para a elaboração do conjunto de testes. Para que HYCONES II fosse validado, foram construídas 46 versões da base de conhecimento hibrida (BCH) para o domínio de cardiopatias congênitas e 46 versões da BCH para o de nefrologia. Em ambos os domínios médicos as respectivas bases de conhecimento foram construídas, automaticamente, a partir das respectivas bases de casos de treinamento. Das 46 versões geradas para cada grupo, uma representa o modelo MNC e 45 os modelos ART. As versões ART dividem-se em grupos de 3: 15 versões foram formadas a partir do modelo Simplified Fuzzy ARTMAP; 15 a partir deste mesmo modelo, sem que os padrões de entrada fossem normalizados; e, finalmente, 15 para o modelo Semantic ART. Na base de testes CHD, o desempenho da versa° HYCONES II - Simplified Fuzzy ARTMAP foi semelhante ao da versa° MNC. A primeira acertou 29 dos 33 diagnósticos (87,9%), enquanto a segunda apontou corretamente 31 dos 33 diagnósticos apresentados (93,9%). Na base de testes de síndromes renais, o desempenho de HYCONES II Fuzzy ARTMAP foi superior ao da versão MNC (p < 0,05). Ambas -Simplified acertaram, respectivamente, 108 (85%) e 95 (74,8%) diagnósticos, em 127 casos submetidos. Ainda que o desempenho da versão HYCONES II - Simplified Fuzzy ARTMAP se revelasse promissor, ao se examinar o conteúdo das redes geradas por este modelo, pode-se observar que estas divergiam completamente daquelas obtidas pelo MNC. As redes que levaram a conclusão diagnostica, na versão HYCONES - MNC, possuíam conteúdo praticamente igual aos grafos de conhecimento, elicitados de especialistas em cardiopatias congênitas. JA, as redes ativadas na versa° HYCONES II - Simplified Fuzzy ARTMAP, além de representarem numero bem major de evidencias que as redes MNC, a grande maioria destas ultimas representam a negação do padrão de entrada. Este fato deve-se a um processo de normalização, inerente ao modelo Simplified Fuzzy ARTMAP, no qual cada padrão de entrada e duplicado. Nesta duplicação, são representadas as evidências presentes em cada caso e, ao mesmo tempo, complementarmente, as evidencias ausentes, em relação ao total geral das mesmas na base de casos. Esta codificação inviabiliza o mecanismo de explanação do sistema HYCONES, pois, na área módica, os diagnósticos costumam ser feitos a partir de um conjunto de evidencias presentes e, não, pela ausência delas. Tentou-se, então, melhorar o conteúdo semântico das redes Simplified Fuzzy ARTMAP. Para tal, o processo de normalização ou codificação complementar da implementação do modelo foi retirado, validando-o novamente, contra o mesma base de testes. Na base de testes CHD, o desempenho de HYCONES II - Simplified Fuzzy ARTMAP, sem a codificação complementar, foi inferior ao da versão MNC (p < 0,05). A primeira acertou 25 dos 33 diagnósticos (75,8%), enquanto a segunda apontou corretamente 31 dos mesmos (93,9%). Na base de testes renais, o desempenho da versa° HYCONES II - Simplified Fuzzy ARTMAP, sem a codificação complementar, foi semelhante ao da versa° MNC. Dos 127 casos apresentados, a primeira acertou 98 diagn6sticos (77,2%), contra 95 da segunda (74,8%). Constatou-se, ainda, que as categorias de reconhecimento formadas pelo modelo Simplified Fuzzy ARTMAP continuavam a apresentar diferenças marcantes quanto ao seu conteúdo, quando comparadas as redes MNC ou aos grafos de conhecimento elicitados de especialistas. O modelo Semantic ART foi, então, proposto, na tentativa de se melhorar o conteúdo semantic° das redes ART. Modificou-se, então, o algoritmo de aprendizado do modelo Simplified Fuzzy ARTMAP, introduzindo-se o mecanismo de aprendizado indutivo do modelo MNC, i.e., o algoritmo de punições e recompensas, associado ao de poda e normalização. Nova validação com a mesma base de testes foi realizada. Para a base de testes de CHD, o desempenho de HYCONES II - SMART foi semelhante ao da versão Simplified Fuzzy ARTMAP e da versão MNC. A primeira e a segunda acertaram 29 dos 33 diagnósticos (87,9%), enquanto a versão MNC apontou corretamente 31 dos 33 diagnósticos apresentados (93,9%). Na base de testes de síndromes renais, o desempenho de HYCONES II - SMART foi superior ao da versão MNC (p < 0,05) e igual ao da versão Simplified Fuzzy ARTMAP. A primeira e a Ultima acertaram 108 dos 127 diagnósticos (85%), enquanto a segunda apontou corretamente 95 dos mesmos (74,8%). Desta feita, observou-se que as redes neurais geradas por HYCONES II - SMART eram semelhantes em conteúdo as redes MNC e aos grafos de conhecimento elicitados de múltiplos especialistas. As principais contribuições desta dissertação são: o projeto, implementação e validação dos modelos Simplified Fuzzy ARTMAP e SMART. Destaca-se, porem, o modelo SMART, que apresentou major valor semântico nas categorias de reconhecimento do que o observado nos modelos ART convencionais, graças a incorporação dos conceitos de especificidade e relevância. Esta dissertação, entretanto, representa não só a modelagem e validação de dois novos modelos neurais, mas sim, o enriquecimento do sistema HYCONES, a partir da continuação de dissertação de mestrado previamente defendida. A partir do presente trabalho, portanto, é dada a possibilidade de escolha, ao engenheiro de conhecimento, de um entre três modelos neurais: o MNC, o Semantic ART e o Simplified Fuzzy ARTMAP que, sem exceção, apresentam Born desempenho. Os dois primeiros destacam-se, contudo, por suportarem semanticamente o contexto. / This dissertation presents two new connectionist models based on the adaptive resonance theory (ART): Simplified Fuzzy ARTMAP and Semantic ART (SMART). The modeling, adaptation, implementation and validation of these models are described, in their association to HYCONES, a hybrid connectionist expert system to solve classification problems. HYCONES integrates the knowledge representation mechanism of frames with neural networks, incorporating the inherent qualities of the two paradigms. While the frames mechanism provides flexible constructs for modeling the domain knowledge, neural networks, implemented in HYCONES' first version by the combinatorial neuron model (CNM), provide the means for automatic knowledge acquisition from a case database, enabling, as well, the implementation of deductive and inductive learning. The Adaptive Resonance Theory (ART) deals with a system involving selfstabilizing input patterns into recognition categories, while maintaining a balance between the properties of plasticity and stability. ART includes a series of different connectionist models: Fuzzy ARTMAP, Fuzzy ART, ART 1, ART 2, and ART 3. Among them, the Fuzzy ARTMAP one stands out for being capable of learning analogical patterns, using two basic ART modules. The Simplified Fuzzy ARTMAP model is a simplification of the Fuzzy ARTMAP neural network. Constrating the first model, the new one is capable of learning analogical patterns using only one ART module. This module is responsible for the categorization of the input patterns. However, it has one more layer, which is responsible for receiving and propagating the target patterns through the network. The presence of a single ART module does not hamper the Simplified Fuzzy ARTMAP model. The same performance levels are attained when the latter one runs without the second ART module. This is certified by the match-tracking strategy, that conjointly maximizes generalization and minimizes predictive error. Two medical domains were chosen to validate HYCONES performance: congenital heart diseases (CHD) and renal syndromes. To build up the CHD case base, 66 medical records were extracted from the cardiac surgery database of the Institute of Cardiology RS (ICFUC-RS). These records cover the period from January 1986 to December 1990 and describe 22 cases of Atrial Septal Defect (ASD), 29 of Ventriculal Septal Defect (VSD), and 15 of Atrial- Ventricular Septa! Defect (AVSD), the three most frequent congenital heart diseases. For validation purposes, 33 additional cases, from the same database and period mentioned above, were also extracted. From these cases, 13 report ASD, 10 VSD and 10 AVSD. To build the renal syndromes case base, 381 medical records from the database of the Escola Paulista de Medicina were analyzed and 58 evidences, covering the patients' clinical history and physical examination data, were semiautomatically extracted. From the total number of selected cases, 136 exhibit Uremia, 85 Nephritis, 100 Hypertension, and 60 Calculosis. From the 381 cases analyzed, 245 were randomically chosen to build the training set, while the remaining ones were used to build the testing set. To validate HYCONES II, 46 versions of the hybrid knowledge base (HKB) with congenital heart diseases were built; for the renal domain, another set of 46 HKB versions were constructed. For both medical domains, the HKBs were automatically generated from the training databases. From these 46 versions, one operates with the CNM model and the other 45 deals with two ART models. These ART versions are divided in three groups: 15 versions were built using the Simplified Fuzzy ARTMAP model; 15 used the Simplified Fuzzy ARTMAP model without the normalization of the input patterns, and 15 used the Semantic ART model. HYCONES II - Simplified Fuzzy ARTMAP and HYCONES - CNM performed similarly for the CH D domain. The first one pointed out correctly to 29 of the 33 testing cases (87,9%), while the second one indicated correctly 31 of the same cases (93,9%). In the renal syndromes domain, however, the performance of HYCONES II - Simplified Fuzzy ARTMAP was superior to the one exhibited by CNM (p < 0,05). Both versions pointed out correctly, respectively, 108 (85%) and 95 (74.8%) diagnoses of the 127 testing cases presented to the system. HYCONES II - Simplified Fuzzy ARTMAP, therefore, displayed a satisfactory performance. However, the semantic contents of the neural nets it generated were completely different from the ones stemming from the CNM version. The networks that pointed out the final diagnosis in HYCONES - CNM were very similar to the knowledge graphs elicited from experts in congenital heart diseases. On the other hand, the networks activated in HYCONES II - Simplified Fuzzy ARTMAP operated with far more evidences than the CNM version. Besides this quantitative difference, there was a striking qualitative discrepancy among these two models. The Simplified Fuzzy ARTMAP version, even though pointing out to the correct diagnoses, used evidences that represented the complementary coding of the input pattern. This coding, inherent to the Simplified Fuzzy ARTMAP model, duplicates the input pattern, generating a new one depicting the evidence observed (on-cell) and, at the same time, the absent evidence, in relation to the total evidence employed to represent the input cases (off-cell). This coding shuts out the HYCONES explanation mechanism, since medical doctors usually reach a diagnostic conclusion rather from a set of observed evidences than from their absence. The next step taken was to improve the semantic contents of the Simplified Fuzzy ARTMAP model. To achieve this, the complement coding process was removed and the modified model was, then, revalidated, through the same testing sets as above described. In the CHD domain, the performance of HYCONES II - Simplified Fuzzy ARTMAP, without complementary coding, proved to be inferior to the one presented by CNM (p < 0,05). The first model singled out correctly 25 out of the 33 testing cases (75,8%), while the second one singled out correctly 31 out of the same 33 cases (93,9%). In the renal syndromes domain, the performances of HYCONES II - Simplified Fuzzy ARTMAP, without complementary coding, and HYCONES - CNM were similar. The first pointed out correctly to 98 of the 127 testing cases (77,2%), while the second one pointed out correctly to 95 of the same cases (74.8%). However, the recognition categories formed by this modified Simplified Fuzzy ARTMAP still presented quantitative and qualitative differences in their contents, when compared to the networks activated by CNM and to the knowledge graphs elicited from experts. This discrepancy, although smaller than the one observed in the original Fuzzy ARTMAP model, still restrained HYCONES explanation mechanism. The Semantic ART model (SMART) was, then, proposed. Its goal was to improve the semantic contents of ART recognition categories. To build this new model, the Simplified Fuzzy ARTMAP archictecture was preserved, while its learning algorithm was replaced by the CNM inductive learning mechanism (the punishments and rewards algorithm, associated with the pruning and normalization mechanisms). A new validation phase was, then, performed over the same testing sets. For the CHD domain, the perfomance comparison among SMART, Simplified Fuzzy ARTMAP, and CNM versions showed similar results. The first and the second versions pointed out correctly to 29 of the 33 testing cases (87,9%), while the third one singled out correctly 31 of the same testing cases (93,9%). For the renal syndromes domain, the performance of HYCONES II - SMART was superior to the one presented by the CNM version (p < 0,05), and equal to the performance presented by the Simplified Fuzzy ARTMAP version. SMART and Simplified Fuzzy ARTMAP singled out correctly 108 of the 127 testing cases (85%), while the CNM version pointed out correctly 95 of the same 127 testing cases (74.8%). Finally, it was observed that the neural networks generated by HYCONES II - SMART had a similar content to the networks generated by CNM and to the knowledge graphs elicited from multiple experts. The main contributions of this dissertation are: the design, implementation and validation of the Simplified Fuzzy ARTMAP and SMART models. The latter one, however, stands out for its learning mechanism, which provides a higher semantic value to the recognition categories, when compared to the categories formed by conventional ART models. This important enhancement is obtained by incorporating specificity and relevance concepts to ART's dynamics. This dissertation, however, represents not only the design and validation of two new connectionist models, but also, the enrichment of HYCONES. This is obtained through the continuation of a previous MSc dissertation, under the same supervision supervision. From the present work, therefore, it is given to the knowledge engineering, the choice among three different neural networks: CNM, Semantic ART and Simplified Fuzzy ARTMAP, all of which, display good performance. Indeed, the first and second models, in contrast to the third, support the context in a semantic way.

O presente trabalho apresenta dois novas modelos conexionistas, baseados na teoria da adaptação ressonante (ART): Simplified Fuzzy ARTMAP e Semantic ART (SMART). Descreve-se a modelagem, adaptação, implementação e validação destes, enquanto incorporados ao sistema hibrido HYCONES, para resolução de problemas de diagnostico medico em cardiopatias congênitas e nefrologia. HYCONES é uma ferramenta para a construção de sistemas especialistas híbridos que integra redes neurais com frames, assimilando as qualidades inerentes aos dois paradigmas. 0 mecanismo de frames fornece tipos construtores flexíveis para a modelagem do conhecimento do domínio, enquanto as redes neurais, representadas na versão original de HYCONES pelo modelo neural combinatório (MNC), possibilitam tanto a automação da aquisição de conhecimento, a partir de uma base de casos, quanta a implementação de aprendizado indutivo e dedutivo. A teoria da adaptação ressonante 6 caracterizada, principalmente, pela manutenção do equilíbrio entre as propriedades de plasticidade e estabilidade durante o processo de aprendizagem. ART inclui vários modelos conexionistas, tais como: Fuzzy ARTMAP, Fuzzy ART, ART 1, ART 2 e ART 3. Dentre estes, a rede neural Fuzzy ARTMAP destaca-se por possibilitar o tratamento de padr6es analógicos a partir de dois módulos ART básicos. O modelo Simplified Fuzzy ARTMAP, como o pr6prio nome o diz, a uma simplificação da rede neural Fuzzy ARTMAP. Ao contrario desta, o novo modelo possibilita o tratamento de padrões analógicos, a partir de apenas um modulo ART, responsável pelo tratamento dos padrões de entrada, adicionado de uma camada, responsável pelos padrões alvo. Mesmo com apenas um modulo ART, o modelo Simplified Fuzzy ARTMAP 6 capaz de reter o mesmo nível de desempenho obtido com a rede neural Fuzzy ARTMAP pois, continua a garantir, conjuntamente, a maximização da generalização e a minimização do erro preditivo, através da execução da estratégia match-tracking. Para a construção da base de casos de cardiopatias congênitas, 66 prontuários médicos, das três cardiopatias congênitas mais freqüentes, foram extraídos do banco de dados de pacientes submetidos a cirurgia cardíaca no Instituto de Cardiologia RS (ICFUC-RS). Tais prontuários abrangem o período de janeiro de 1986 a dezembro de 1990 e reportam 22 casos de Comunicação Interatrial (CIA), 29 de Comunicação Interventricular (CIV) e 15 de Defeito Septal Atrioventricular (DSAV). Para a análise de desempenho do sistema, 33 casos adicionais, do referido período, foram extraídos aleatoriamente do banco de dados do ICFUC-RS. Destes 33 casos, 13 apresentam CIA, 10 CIV e 10 DSAV. Para a construção da base de casos de síndromes renais, 381 prontuários do banco de dados de síndromes renais da Escola Paulista de Medicina foram analisados e 58 evidencias, correspondentes a dados de hist6ria clinica e exame físico dos pacientes, foram extraídas semi-automaticamente. Do total de casos selecionados, 136 apresentam Uremia, 85 Nefrite, 100 Hipertensão e 60 Litiase. Dos 381 casos analisados, 254 foram escolhidos aleatoriamente para a composicao do conjunto de treinamento, enquanto que os demais foram utilizados para a elaboração do conjunto de testes. Para que HYCONES II fosse validado, foram construídas 46 versões da base de conhecimento hibrida (BCH) para o domínio de cardiopatias congênitas e 46 versões da BCH para o de nefrologia. Em ambos os domínios médicos as respectivas bases de conhecimento foram construídas, automaticamente, a partir das respectivas bases de casos de treinamento. Das 46 versões geradas para cada grupo, uma representa o modelo MNC e 45 os modelos ART. As versões ART dividem-se em grupos de 3: 15 versões foram formadas a partir do modelo Simplified Fuzzy ARTMAP; 15 a partir deste mesmo modelo, sem que os padrões de entrada fossem normalizados; e, finalmente, 15 para o modelo Semantic ART. Na base de testes CHD, o desempenho da versa° HYCONES II - Simplified Fuzzy ARTMAP foi semelhante ao da versa° MNC. A primeira acertou 29 dos 33 diagnósticos (87,9%), enquanto a segunda apontou corretamente 31 dos 33 diagnósticos apresentados (93,9%). Na base de testes de síndromes renais, o desempenho de HYCONES II Fuzzy ARTMAP foi superior ao da versão MNC (p < 0,05). Ambas -Simplified acertaram, respectivamente, 108 (85%) e 95 (74,8%) diagnósticos, em 127 casos submetidos. Ainda que o desempenho da versão HYCONES II - Simplified Fuzzy ARTMAP se revelasse promissor, ao se examinar o conteúdo das redes geradas por este modelo, pode-se observar que estas divergiam completamente daquelas obtidas pelo MNC. As redes que levaram a conclusão diagnostica, na versão HYCONES - MNC, possuíam conteúdo praticamente igual aos grafos de conhecimento, elicitados de especialistas em cardiopatias congênitas. JA, as redes ativadas na versa° HYCONES II - Simplified Fuzzy ARTMAP, além de representarem numero bem major de evidencias que as redes MNC, a grande maioria destas ultimas representam a negação do padrão de entrada. Este fato deve-se a um processo de normalização, inerente ao modelo Simplified Fuzzy ARTMAP, no qual cada padrão de entrada e duplicado. Nesta duplicação, são representadas as evidências presentes em cada caso e, ao mesmo tempo, complementarmente, as evidencias ausentes, em relação ao total geral das mesmas na base de casos. Esta codificação inviabiliza o mecanismo de explanação do sistema HYCONES, pois, na área módica, os diagnósticos costumam ser feitos a partir de um conjunto de evidencias presentes e, não, pela ausência delas. Tentou-se, então, melhorar o conteúdo semântico das redes Simplified Fuzzy ARTMAP. Para tal, o processo de normalização ou codificação complementar da implementação do modelo foi retirado, validando-o novamente, contra o mesma base de testes. Na base de testes CHD, o desempenho de HYCONES II - Simplified Fuzzy ARTMAP, sem a codificação complementar, foi inferior ao da versão MNC (p < 0,05). A primeira acertou 25 dos 33 diagnósticos (75,8%), enquanto a segunda apontou corretamente 31 dos mesmos (93,9%). Na base de testes renais, o desempenho da versa° HYCONES II - Simplified Fuzzy ARTMAP, sem a codificação complementar, foi semelhante ao da versa° MNC. Dos 127 casos apresentados, a primeira acertou 98 diagn6sticos (77,2%), contra 95 da segunda (74,8%). Constatou-se, ainda, que as categorias de reconhecimento formadas pelo modelo Simplified Fuzzy ARTMAP continuavam a apresentar diferenças marcantes quanto ao seu conteúdo, quando comparadas as redes MNC ou aos grafos de conhecimento elicitados de especialistas. O modelo Semantic ART foi, então, proposto, na tentativa de se melhorar o conteúdo semantic° das redes ART. Modificou-se, então, o algoritmo de aprendizado do modelo Simplified Fuzzy ARTMAP, introduzindo-se o mecanismo de aprendizado indutivo do modelo MNC, i.e., o algoritmo de punições e recompensas, associado ao de poda e normalização. Nova validação com a mesma base de testes foi realizada. Para a base de testes de CHD, o desempenho de HYCONES II - SMART foi semelhante ao da versão Simplified Fuzzy ARTMAP e da versão MNC. A primeira e a segunda acertaram 29 dos 33 diagnósticos (87,9%), enquanto a versão MNC apontou corretamente 31 dos 33 diagnósticos apresentados (93,9%). Na base de testes de síndromes renais, o desempenho de HYCONES II - SMART foi superior ao da versão MNC (p < 0,05) e igual ao da versão Simplified Fuzzy ARTMAP. A primeira e a Ultima acertaram 108 dos 127 diagnósticos (85%), enquanto a segunda apontou corretamente 95 dos mesmos (74,8%). Desta feita, observou-se que as redes neurais geradas por HYCONES II - SMART eram semelhantes em conteúdo as redes MNC e aos grafos de conhecimento elicitados de múltiplos especialistas. As principais contribuições desta dissertação são: o projeto, implementação e validação dos modelos Simplified Fuzzy ARTMAP e SMART. Destaca-se, porem, o modelo SMART, que apresentou major valor semântico nas categorias de reconhecimento do que o observado nos modelos ART convencionais, graças a incorporação dos conceitos de especificidade e relevância. Esta dissertação, entretanto, representa não só a modelagem e validação de dois novos modelos neurais, mas sim, o enriquecimento do sistema HYCONES, a partir da continuação de dissertação de mestrado previamente defendida. A partir do presente trabalho, portanto, é dada a possibilidade de escolha, ao engenheiro de conhecimento, de um entre três modelos neurais: o MNC, o Semantic ART e o Simplified Fuzzy ARTMAP que, sem exceção, apresentam Born desempenho. Os dois primeiros destacam-se, contudo, por suportarem semanticamente o contexto. / This dissertation presents two new connectionist models based on the adaptive resonance theory (ART): Simplified Fuzzy ARTMAP and Semantic ART (SMART). The modeling, adaptation, implementation and validation of these models are described, in their association to HYCONES, a hybrid connectionist expert system to solve classification problems. HYCONES integrates the knowledge representation mechanism of frames with neural networks, incorporating the inherent qualities of the two paradigms. While the frames mechanism provides flexible constructs for modeling the domain knowledge, neural networks, implemented in HYCONES' first version by the combinatorial neuron model (CNM), provide the means for automatic knowledge acquisition from a case database, enabling, as well, the implementation of deductive and inductive learning. The Adaptive Resonance Theory (ART) deals with a system involving selfstabilizing input patterns into recognition categories, while maintaining a balance between the properties of plasticity and stability. ART includes a series of different connectionist models: Fuzzy ARTMAP, Fuzzy ART, ART 1, ART 2, and ART 3. Among them, the Fuzzy ARTMAP one stands out for being capable of learning analogical patterns, using two basic ART modules. The Simplified Fuzzy ARTMAP model is a simplification of the Fuzzy ARTMAP neural network. Constrating the first model, the new one is capable of learning analogical patterns using only one ART module. This module is responsible for the categorization of the input patterns. However, it has one more layer, which is responsible for receiving and propagating the target patterns through the network. The presence of a single ART module does not hamper the Simplified Fuzzy ARTMAP model. The same performance levels are attained when the latter one runs without the second ART module. This is certified by the match-tracking strategy, that conjointly maximizes generalization and minimizes predictive error. Two medical domains were chosen to validate HYCONES performance: congenital heart diseases (CHD) and renal syndromes. To build up the CHD case base, 66 medical records were extracted from the cardiac surgery database of the Institute of Cardiology RS (ICFUC-RS). These records cover the period from January 1986 to December 1990 and describe 22 cases of Atrial Septal Defect (ASD), 29 of Ventriculal Septal Defect (VSD), and 15 of Atrial- Ventricular Septa! Defect (AVSD), the three most frequent congenital heart diseases. For validation purposes, 33 additional cases, from the same database and period mentioned above, were also extracted. From these cases, 13 report ASD, 10 VSD and 10 AVSD. To build the renal syndromes case base, 381 medical records from the database of the Escola Paulista de Medicina were analyzed and 58 evidences, covering the patients' clinical history and physical examination data, were semiautomatically extracted. From the total number of selected cases, 136 exhibit Uremia, 85 Nephritis, 100 Hypertension, and 60 Calculosis. From the 381 cases analyzed, 245 were randomically chosen to build the training set, while the remaining ones were used to build the testing set. To validate HYCONES II, 46 versions of the hybrid knowledge base (HKB) with congenital heart diseases were built; for the renal domain, another set of 46 HKB versions were constructed. For both medical domains, the HKBs were automatically generated from the training databases. From these 46 versions, one operates with the CNM model and the other 45 deals with two ART models. These ART versions are divided in three groups: 15 versions were built using the Simplified Fuzzy ARTMAP model; 15 used the Simplified Fuzzy ARTMAP model without the normalization of the input patterns, and 15 used the Semantic ART model. HYCONES II - Simplified Fuzzy ARTMAP and HYCONES - CNM performed similarly for the CH D domain. The first one pointed out correctly to 29 of the 33 testing cases (87,9%), while the second one indicated correctly 31 of the same cases (93,9%). In the renal syndromes domain, however, the performance of HYCONES II - Simplified Fuzzy ARTMAP was superior to the one exhibited by CNM (p < 0,05). Both versions pointed out correctly, respectively, 108 (85%) and 95 (74.8%) diagnoses of the 127 testing cases presented to the system. HYCONES II - Simplified Fuzzy ARTMAP, therefore, displayed a satisfactory performance. However, the semantic contents of the neural nets it generated were completely different from the ones stemming from the CNM version. The networks that pointed out the final diagnosis in HYCONES - CNM were very similar to the knowledge graphs elicited from experts in congenital heart diseases. On the other hand, the networks activated in HYCONES II - Simplified Fuzzy ARTMAP operated with far more evidences than the CNM version. Besides this quantitative difference, there was a striking qualitative discrepancy among these two models. The Simplified Fuzzy ARTMAP version, even though pointing out to the correct diagnoses, used evidences that represented the complementary coding of the input pattern. This coding, inherent to the Simplified Fuzzy ARTMAP model, duplicates the input pattern, generating a new one depicting the evidence observed (on-cell) and, at the same time, the absent evidence, in relation to the total evidence employed to represent the input cases (off-cell). This coding shuts out the HYCONES explanation mechanism, since medical doctors usually reach a diagnostic conclusion rather from a set of observed evidences than from their absence. The next step taken was to improve the semantic contents of the Simplified Fuzzy ARTMAP model. To achieve this, the complement coding process was removed and the modified model was, then, revalidated, through the same testing sets as above described. In the CHD domain, the performance of HYCONES II - Simplified Fuzzy ARTMAP, without complementary coding, proved to be inferior to the one presented by CNM (p < 0,05). The first model singled out correctly 25 out of the 33 testing cases (75,8%), while the second one singled out correctly 31 out of the same 33 cases (93,9%). In the renal syndromes domain, the performances of HYCONES II - Simplified Fuzzy ARTMAP, without complementary coding, and HYCONES - CNM were similar. The first pointed out correctly to 98 of the 127 testing cases (77,2%), while the second one pointed out correctly to 95 of the same cases (74.8%). However, the recognition categories formed by this modified Simplified Fuzzy ARTMAP still presented quantitative and qualitative differences in their contents, when compared to the networks activated by CNM and to the knowledge graphs elicited from experts. This discrepancy, although smaller than the one observed in the original Fuzzy ARTMAP model, still restrained HYCONES explanation mechanism. The Semantic ART model (SMART) was, then, proposed. Its goal was to improve the semantic contents of ART recognition categories. To build this new model, the Simplified Fuzzy ARTMAP archictecture was preserved, while its learning algorithm was replaced by the CNM inductive learning mechanism (the punishments and rewards algorithm, associated with the pruning and normalization mechanisms). A new validation phase was, then, performed over the same testing sets. For the CHD domain, the perfomance comparison among SMART, Simplified Fuzzy ARTMAP, and CNM versions showed similar results. The first and the second versions pointed out correctly to 29 of the 33 testing cases (87,9%), while the third one singled out correctly 31 of the same testing cases (93,9%). For the renal syndromes domain, the performance of HYCONES II - SMART was superior to the one presented by the CNM version (p < 0,05), and equal to the performance presented by the Simplified Fuzzy ARTMAP version. SMART and Simplified Fuzzy ARTMAP singled out correctly 108 of the 127 testing cases (85%), while the CNM version pointed out correctly 95 of the same 127 testing cases (74.8%). Finally, it was observed that the neural networks generated by HYCONES II - SMART had a similar content to the networks generated by CNM and to the knowledge graphs elicited from multiple experts. The main contributions of this dissertation are: the design, implementation and validation of the Simplified Fuzzy ARTMAP and SMART models. The latter one, however, stands out for its learning mechanism, which provides a higher semantic value to the recognition categories, when compared to the categories formed by conventional ART models. This important enhancement is obtained by incorporating specificity and relevance concepts to ART's dynamics. This dissertation, however, represents not only the design and validation of two new connectionist models, but also, the enrichment of HYCONES. This is obtained through the continuation of a previous MSc dissertation, under the same supervision supervision. From the present work, therefore, it is given to the knowledge engineering, the choice among three different neural networks: CNM, Semantic ART and Simplified Fuzzy ARTMAP, all of which, display good performance. Indeed, the first and second models, in contrast to the third, support the context in a semantic way.

Cette thèse traite des systèmes de recommandation automatiques. Les moteurs de recommandation automatique sont des systèmes qui permettent, par des techniques de data mining, de recommander automatiquement à des clients, en fonction de leurs consommations passées, des produits susceptibles de les intéresser. Ces systèmes permettent par exemple d'augmenter les ventes sur des sites web marchands : le site Amazon a une stratégie marketing en grande partie basée sur la recommandation automatique. Amazon a popularisé l'usage de la recommandation automatique par la célèbre fonction de recommandation que nous qualifions d'item-to-items, le fameux : " les personnes qui ont vu/acheté cet articles ont aussi vu/acheté ces articles. La contribution centrale de cette thèse est d'analyser les systèmes de recommandation automatiques dans le contexte industriel, et notamment des besoins marketing, et de croiser cette analyse avec les travaux académiques. / This thesis deals with automatic recommendation systems. Automatic recommendation systems are systems that allow, through data mining techniques, to recommend automatically to users, based on their past consumption, items that may interest them. These systems allow for example to increase sales on e-commerce websites: the Amazon site has a marketing strategy based mainly on the recommendation. Amazon has popularized the use of automatic recommendation based on the recommendation function that we call item-to-items, the famous "people who have seen / bought this product have also seen / bought these articles". The central contribution of this thesis is to analyze the automatic recommendation systems in the industrial context, including marketing needs, and to cross this analysis with academic works.

Ce travail de thèse s’inscrit dans la recherche des solutions économiquementviables pour des réseaux personnels à hauts débits (plusieurs Gbps à plusieursdizaines de Gbps) opérationnels en bande millimétrique autour de 60 GHz. Aucas où ces réseaux servent un nombre élevé d’utilisateurs, ils comprendront unemultitude d’antennes afin d’assurer l’accès sans fil rapide. Afin de réduire aumaximum le coût d’un module d’antenne, les réseaux doivent fournir un signalanalogue à des porteuses millimetriques. Une solution prometteuse pour les systèmesde distribution qui correspond à ces besoins sont des structures à fibreoptique, laquelle permet une transmission à faibles pertes et à haute bande passante.On parle de l’approche "radio-sur-fibre" (en anglais, radio-over-fiber). Laproblématique est de pouvoir générer et moduler un signal aux fréquences millimétriqueslors de la transmission optique - et ce avec des composant bas coûts.La technique utilisée dans le cadre de cette thèse est l’emploi des diodes laser àverrouillage de modes. Ces derniers vont pouvoir générer des hautes fréquencestout en ne nécessitant qu’une alimentation continue, et ils peuvent être modulésde manière directe ou externe. Les lasers à semi-conducteurs employés ici sontd’une génération encore à l’état d’étude puisqu’il s’agit des lasers à boites (ouîlots) quantiques. Ces lasers ont montrés de très bonnes capacités à générer dessignaux électriques aux fréquences autour de 60 GHz, bien qu’ayant encore, pourl’instant, à une stabilité de fréquence (ou de phase) limitée. Dans le cadre des systèmesde communication opto/micro-ondes, peu de travaux approfondis ont étémenés sur ces structures.Au cours de cette thèse, plusieurs études ont été effectuées. La première portesur les propriétés générales d’un système construit à partir de ce type de laser(puissances disponibles, figure de bruit, linéarité etc.). Une deuxième étude aété consacrée aux effets de la propagation des signaux dans les systèmes baséssur les lasers à verrouillage de modes, notamment de la dispersion chromatiquelaquelle a un effet considérable sur les distances de transmission. Les deux étudesmettent en avant l’importance d’une limitation du nombre de modes générés parla diode laser afin d’optimiser non seulement le gain du lien et la puissance RFrécupérée, mais aussi la figure de bruit du système. Lors d’une troisième étude, lastabilité en fréquence/phase s’est révélée critique, car le bruit de fréquence/phaselimite la qualité de la transmission en introduisant un plancher d’erreur mêmepour des rapports signal-a-bruit très élevés. Des différentes générations de lasersà boites (îlots) quantiques et à verrouillage de modes ont été testées. Le problèmedu bruit de fréquence et de phase persiste et ne peut pas être résolu en utilisantles techniques classiques comme les boucles à verrouillage de phase conventionnelles.Une solution pour ce problème a été développée pour les systèmes detransmission; elle permet simultanément un ajustement de fréquence supérieure(précision de quelques Hz à quelques kHz) à celle donnée par le processus de fabricationdes diodes lasers (précision de quelques GHz), ainsi qu’une stabilisationde fréquence et de phase. / This dissertation is related to the search for an economically sustainable solutionfor high data rate (several Gbps to several tens of Gbps) personal area networksoperating in the millimeter-wave region around 60 GHz. If such networks supplya large number of users, they need to encompass a multitude of antenna pointsin order to assure wireless access to the network. With the aim of reducing thecost of an antenna module, the networks should at best provide quasi "readyto-radiate" signals to the modules, i.e. at millimeter-wave carrier frequencies.Thanks to their low transmission loss and their high bandwidth, optical fiber distributionarchitectures represent a promising solution. The technique is referredto as the so-called "radio-over-fiber" approach whereby the analog radio signalwill be transported to the access point by an optical wave. The challenge herebyis the generation and modulation of an optical signal by a millimeter-wave radiosignal using preferably cost-efficient system components. The technique proposedherein is based on the use of mode-locked laser diodes which can generatesignals at very high frequencies under the condition of continuous current supply.Mode-locked laser diodes can be modulated both directly and externally. Thediodes employed in this work are based on so-called quantum dots (or quantumdashes); these are material structures which are themselves still subject to intensivephysical research. Signals at millimeter-wave frequencies (around 60 GHz)can easily be generated by such lasers. However, their frequency and phase stabilityis as yet limited. In the context of radio-over-fiber communication systems,these structures have not yet been studied in detail.In the course of this dissertation, several aspects are considered. A first systemstudy treats the basic properties of a system built from this type of laser source(available signal power, system noise figure, linearity etc.). A second study isdevoted to an investigation of propagation effects like dispersion, which considerablyinfluence the attainable transmission distances. An essential result of bothstudies is the importance of limiting the laser spectrum to a small number of lasermodes for an optimization of link gain, generated RF power, and system noisefigure. A third study deals with the limited frequency and phase stability whichturn out to be critical factors for transmission quality. The study of several generationsof quantum dot/dash lasers has revealed that the problems of frequencyand phase noise persist and cannot be solved using classical techniques involvinge.g. conventional phase-locked loops. In this dissertation, a solution is presentedwhich not only allows a more precise adjustment of the laser frequency (precisionin the order of Hz to kHz) than that given by the manufacturing process of thelaser (precision in the order of GHz), but also enables a stabilization of frequencyand phase. / Die vorliegende Dissertation steht im Zusammenhang mit der Suche nach wirtschaftlichtragfähigen Lösungen zum Aufbau hochdatenratiger Heimnetzwerke(einige Gbps bis einige zehn Gbps), so genannter Personal area-Netzwerke imMillimeterwellenbereich um 60 GHz. Sollen diese Netze eine große Anzahl vonNutzern versorgen, wird eine Vielzahl von Zugangspunkten - also Antennenmodulen- benötigt, um den drahtlosen Netzanschluss zu ermöglichen. Um dieKosten eines Antennenmoduls soweit wie möglich zu senken, sollen die Netzequasi "abstrahlfertige" Signale an die Module liefern, d. h. auf Trägerfrequenzenim Millimeterwellenbereich. Glasfaserbasierte Verteilsysteme werden dankihrer geringen Leitungsverluste und ihrer hohen Bandbreite diesem Anspruchgerecht. Man spricht hier vom so genannten Radio-over-fiber-Ansatz, wobei dasanaloge Signal von einer optischen Welle zum Zugangspunkt transportiert wird.Die Herausforderung liegt hierbei in der Generierung und Modulation eines optischenSignals mit einem Nutzsignal imMillimeterwellenbereich - und das unterVerwendung möglichst kostengünstiger Komponenten. Die hier vorgeschlageneTechnik basiert auf der Nutzung von modengekoppelten Laserdioden, welcheallein bei Gleichstromversorgung Signale bei hohen Frequenzen erzeugen undsowohl direkt als auch extern moduliert werden können. Die Dioden, welche hierzur Verwendung kommen, basieren auf so genannten Quantenpunkten (englisch:quantum dot/quantum dash); es sind Strukturen, die selbst noch Gegenstand intensiverphysikalischer Forschung sind. Signale bei Frequenzen um 60 GHz könnenleicht von diesen Lasern erzeugt werden, wenn auch bisher nur bei begrenzterFrequenz- und Phasenstabilität. Im Kontext von Radio-over-fiber-Systemenwurden diese Strukturen noch nicht untersucht.Im Rahmen dieser Dissertation wurden mehrere Aspekte betrachtet. Eine ersteSystemstudie behandelt die grundlegendenEigenschaften eines Systems, welchesauf dieser Art von Lasern basiert (verfügbare Leistung, Rauschzahl, Linearitätusw.). Eine zweite Untersuchung ist der Erforschung von Ausbreitungseffektenwie etwa Dispersion gewidmet, welche die erreichbaren Entfernungen maßgeblichbeeinflusst. Ein wesentliches Ergebnis beider Studien ist die Relevanzeiner Begrenzung des Laserspektrums auf wenige Moden zur Optimierung vonGewinn, Hochfrequenz-Leistung und Rauschzahl. Eine dritte Studie untersuchtdie Frequenz-und die Phasenstabilität, welche sich als kritisch für die Übertragungsqualitäterweisen. Die Untersuchung von mehreren Generationen von modengekoppeltenQuantenpunktlasern hat ergeben, dass das Problem des FrequenzundPhasenrauschens fortbesteht und nicht auf konventionellem Weg wie z.B.durch die Verwendung von klassischen Phasenregelkreisen gelöst werden kann.Im Rahmen der Arbeit wurde eine Lösung für dieses Problem gefunden, welcheerstens eine bessere Feineinstellung der Frequenz erlaubt (Genauigkeit von Hzbis kHz), als sie durch den Laserfertigungsprozess gegeben ist (Genauigkeit vonGHz), und zweitens eine Stabilisierung von Frequenz und Phase ermöglicht.

Orientadores: Fernando Jose Von Zuben, Mauricio Fernandes Figueiredo / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-06T19:35:31Z (GMT). No. of bitstreams: 1 Michelan_Roberto_M.pdf: 4495515 bytes, checksum: aed72feefc89070579190e862ea0f740 (MD5) Previous issue date: 2006 / Resumo: A concepção de sistemas autônomos de navegação para robôs móveis, havendo múltiplos objetivos simultâneos a serem atendidos, como a coleta de lixo com manutenção da integridade, requer a adoção de técnicas refinadas de coordenação de módulos de comportamento elementar. Modelos de redes imunológicas artificiais podem então ser empregados na proposição de um controlador concebido com base em um processo de mapeamento dinâmico. Os anticorpos da rede são responsáveis pelos módulos de comportamento elementar, na forma de regras do tipo <condição>-<ação>, e as conexões são responsáveis pelos mecanismos de estímulo e supressão entre os anticorpos. A rede iniciará uma resposta imunológica sempre que lhe forem apresentados os antígenos. Estes antígenos representam a situação atual capturada pelos sensores do robô. A dinâmica da rede é baseada no nível de concentração dos anticorpos, definida com base na interação dos anticorpos e dos anticorpos com os antígenos. De acordo com o nível de concentração, um anticorpo é escolhido para definir a ação do robô. Um processo evolutivo é então responsável por definir um padrão de conexões para a rede imunológica, a partir de uma população de redes candidatas, capaz de maximizar o atendimento dos objetivos durante a navegação. Resulta então um sistema híbrido que tem a rede imunológica como responsável por introduzir um processo dinâmico de tomada de decisão e tem agora a computação evolutiva como responsável por definir a estrutura da rede. Para que fosse possível avaliar os controladores (redes imunológicas) a cada geração do processo evolutivo, um ambiente virtual foi desenvolvido para simulação computacional, com base nas características do problema de navegação. As redes imunológicas obtidas através do processo evolutivo foram analisadas e testadas em novas situações, apresentando capacidade de coordenação em tarefas simples e complexas. Os experimentos preliminares com um robô real do tipo Khepera II indicaram a eficácia da ferramenta de navegação / Abstract: The design of an autonomous navigation system for mobile robots, with simultaneous objectives to be satisfied, as garbage collection with maintenance of integrity, requires refined coordination mechanisms to deal with modules of elementary behavior. Models of artificial immune networks can then be applied to produce a controller based on dynamic mapping. The antibodies of the immune network are responsible for the modules of elementary behavior, in the form of <condition>-<action> rules, and the connections are responsible for the mechanisms of stimulation and suppression of antibodies. The network will always start an immune response when antigens are presented. These antigens represent the current output of the robot sensors. The network dynamics is based on the levels of antibody concentration, provided by interaction among antibodies, and among antibodies and antigens. Based on its concentration level, an antibody is chosen to define the robot action. An evolutionary process is then used to define the connection pattern of the immune network, from a population of candidate networks, capable of maximizing the objectives during navigation. As a consequence, a hybrid system is conceived, with an immune network implementing a dynamic process of decision-making, and an evolutionary algorithm defining the network structure. To be able to evaluate the controllers (immune networks) at each iteration of the evolutionary process, a virtual environment was developed for computer simulation, based on the characteristics of the navigation problem. The immune networks obtained by evolution were analyzed and tested in new situations and presented coordination capability in simple and complex tasks. The preliminary experiments on a real Khepera II robot indicated the efficacy of the navigation tool / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Motivée par les besoins du domaine spatial en termes de diagnostic embarqué et d’autonomie, cette thèse s’intéresse aux problèmes de diagnostic, de diagnosticabilité et de diagnostic actif des systèmes hybrides. Un formalisme hybride est proposé pour représenter les deux dynamiques, continues et discrètes, du système. En s’appuyant sur ce modèle, une approche de diagnostic passif est proposée en mariant les techniques des systèmes à événements discrets et des systèmes continus. Un cadre formel pour la diagnosticabilité des systèmes hybrides a également été établi proposant des définitions et des critères pour la diagnosticabilité hybride. Suite à un diagnostic passif ambigu, le diagnostic actif est nécessaire afin de désambiguïser l’état du système. Cette thèse propose donc une approche de diagnostic actif, qui partant d’un état de croyance incertain, fait appel aux propriétés de diagnosticabilité du système pour déterminer la configuration où les fautes peuvent être discriminées. Une nouvelle machine à états finis appelée diagnostiqueur actif est introduite permettant de formaliser le diagnostic actif comme un problème de planification conditionnelle. Un algorithme d’exploration de graphes ET-OU est proposé pour calculer les plans de diagnostic actif. Finalement, l’approche de diagnostic a été testée sur le Système de Contrôle d’Attitude (SCA) d’un satellite de Thales Alenia Space. Le module de diagnostic a été intégré dans la boucle fermée de commande. Des scénarios de faute ont été testés donnant des résultats très satisfaisants. / Motivated by the requirements of the space domain in terms of on-board diagnosis and autonomy, this thesis addresses the problems of diagnosis, diagnosability and active diagnosis of hybrid systems. Supported by a hybrid modeling framework, a passive approach for model-based diagnosis mixing discrete-event and continuous techniques is proposed. The same hybrid model is used to define the diagnosability property for hybrid systems and diagnosability criteria are derived. When the diagnosis provided by the passive diagnosis approach is ambiguous, active diagnosis is needed. This work provides a method for performing such active diagnosis. Starting with an ambiguous belief state, the method calls for diagnosability analysis results to determine a new system configuration in which fault candidates can be discriminated. Based on a new finite state machine called the diagnoser, the active diagnosis is formulated as a conditional planning problem and an AND-OR graph exploration algorithm is proposed to determine active diagnosis plans. Finally, the diagnosis approach is tested on the Attitude Control System (ACS) of a satellite simulator provided by Thales Alenia Space. The diagnosis module is successfully tested on several fault scenarios and the obtained results are reported.

Plusieurs laboratoires de robotique à travers le monde travaillent sur le développement de stratégies innovantes pour la navigation autonome de véhicules élémentaires ou en convoi. Dans ce contexte, nos travaux de thèse s’inscrivent principalement dans le cadre de la navigation en formation d’un groupe de véhicules dans des environnements structurés. La complexité de ces systèmes multi-robots ne permet pas l’utilisation directe de techniques classiques de perception et/ou de contrôle/commande. Nos travaux ont consisté à décomposer le contrôle global, dédié à la réalisation de la tâche complexe, en un ensemble de comportements/contrôleurs élémentaires précis et fiables (e.g., évitement d’obstacles, suivi de trajectoire, attraction vers une cible, navigation en formation, etc.). Ces comportements lient les différentes informations fournies par les capteurs aux actions des véhicules. Pour garantir les critères de performances imposés à notre architecture de contrôle/commande (e.g., stabilité, robustesse et/ou borner les erreurs maximales), les potentialités des systèmes hybrides ont été considérées. Cette architecture de contrôle a été validée, dans un premier temps, sur des véhicules pris individuellement, en utilisant notamment une stratégie de navigation sûre et flexible utilisant des points de passage. Cette navigation permet au véhicule d’effectuer différentes manœuvres entre ces points de passage (pour éviter par exemple des obstacles dans l’environnement) et ce sans avoir à planifier/re-planifier des trajectoires globales dans l’environnement. Une loi de commande spécifique, permettant une attraction stable (au sens de Lyapunov) et précise vers des cibles statiques ou dynamiques a été par ailleurs développée. Cette loi de commande garantit la convergence du véhicule vers chaque point de passage tout en garantissant des trajectoires sûres. Par ailleurs, un algorithme nommé OMWS (pour Optimal Multi-criteria Waypoint Selection) a été proposé pour sélectionner les configurations optimales des points de passage dans l’environnement. Cet algorithme permet de garantir des mouvements sûrs et fiables du véhicules en tenant compte des contraintes et incertitudes liées à la navigation du véhicule. Par la suite, l’architecture de contrôle/commande proposée a été étendue aux systèmes multi-robots en utilisant la combinaison d’une approche leader-suiveur et comportementale. Un important aspect de la navigation multi-robots est la reconfiguration dynamique de la formation en fonction du contexte de la navigation (e.g., passer d’une configuration triangle vers ligne si la largeur de la voie de navigation ne suffisait pas). Ainsi, des stratégies de reconfiguration dynamique ont été proposées, permettant de garantir la sureté de la formation même au moment des transitions entre configurations. Il est à noter par ailleurs que des métriques spécifiques ont été proposées pour quantifier la fiabilité et la robustesse des stratégies multi-robots proposées. Plusieurs simulations et expérimentations avec des véhicules urbains (VIPALABs) nous ont permis de confirmer la viabilité et efficacité des architectures de contrôle/commande proposées pour la navigation en formation d’un groupe de VIPALABs. / Beyond the interest of robotics laboratories for the development of dedicated strategies for single vehicle navigation, several laboratories around the world are more and more involved in the general challenging field of cooperative multi-robot navigation. In this context, this work deals with the navigation in formation of a group of Unmanned Ground Vehicles (UGVs) dedicated to structured environments. The complexity of this Multi-Robot System (MRS) does not permit the direct use of neither classical perception nor control techniques. To overcome this problem, this work proposes to break up the overall control dedicated to the achievement of the complex task into a group of accurate and reliable elementary behaviors/controllers (e.g., obstacles avoidance, trajectory tracking, target reaching, navigation in formation, formation reconfiguration, etc.). These behaviors are linked to different information given by the sensors to the actions of vehicles. To guarantee the performances criteria (e.g., stability, convergence, state errors) aimed by the control architecture, the potentialities of hybrid controllers (which controlling continuous systems in the presence of discrete events) are considered. This control architecture is validated for a single vehicle to perform safe and flexible autonomous navigation using an appropriate strategy of navigation through suitable set of waypoints. This flexible navigation allows different vehicle maneuvers between waypoints (e.g., target reaching or obstacle avoidance) without using any trajectory planning nor replanning. The designed control law based on Lyapunov synthesis guarantees the convergence to assigned waypoint while performing safe trajectories. Furthermore, an algorithm to select suitable waypoints’ positions, named Optimal Multi-criteria Waypoint Selection (OMWS), in structured environments while taking into account the safe and reliable vehicle movements, and vehicle constraints and uncertainties is proposed. Subsequently, the control architecture is extended to Multi-Robot Formation (MRF) using a combination of Leader-Follower and behavior-based approaches. An important cooperative MRS issues in this thesis is the dynamic reconfiguration of the formation according to the context of navigation (e.g., to pass from a triangle configuration towards a line if the width of the navigation way is not sufficient). The proposed Strategy for Formation Reconfiguration (SFR) guarantees the stability and the safety of the MRS at the time of the transitions between configuration (e.g., line towards square, triangle towards line, etc.). Therefore, a safe, reactive and dynamic MRF is obtained. Moreover, the degrees of multi-robot safety, stability and reliability of the system are quantified via suitable metrics. Simulations and experiments using urban vehicles (VIPALABs) of the Institut Pascal laboratory allow to perform exhaustive experiments of the proposed control architecture for the navigation in formation of a group of UGVs.

The conventional approach to validate the analog and mixed signal designs utilizes extensive SPICE-level simulations. The main challenge in this approach is to know when all important corner cases have been simulated. An alternate approach is to use the formal verification techniques. Formal verification techniques have gained wide spread popularity in the digital design domain; but in case of analog and mixed signal designs, a large number of test scenarios need to be designed to generate sufficient simulation traces to test out all the specified system behaviours. Analog and mixed signal designs can be formally modeled as hybrid systems and therefore techniques used for formal analysis and verification of hybrid systems can be applied to the analog and mixed signal designs. Generally, formal verification tools for hybrid systems work at the abstract level where we model the systems in terms of differential equations or algebraic equations. However the analog and mixed signal system designers are very comfortable in designing the circuits at the transistor level. To bridge the gap between abstraction level verification and the designs validation which has been implemented at the transistor level, the very important issue we need to address is: Can we formally verify the circuits at the transistor level itself? For this we have proposed a framework for doing the formal verification of analog and mixed signal designs using SPICE simulation traces in one of the hybrid systems formal verification tools (i.e. Checkmate from CMU). An extension to a formal verification approach of hybrid systems is proposed to verify analog and mixed signal (AMS) designs. AMS designs can be formally modeled as hybrid systems and therefore lend themselves to the formal analysis and verification techniques applied to hybrid systems. The proposed approach employs simulation traces obtained from an actual design implementation of AMS circuit blocks (for example, in the form of SPICE netlists) to carry out formal analysis and verification. This enables the same platform used for formally validating an abstract model of an AMS design to be also used for validating its different refinements and design implementation, thereby providing a simple route to formal verification at different levels of implementation. Our approach has been illustrated through the case studies using simulation traces form the different frameworks i.e. Simulink/Stateflow framework and the SPICE simulation traces. We demonstrate the feasibility of our approach around the Checkmate and the case studies for hybrid systems and the analog and mixed signal designs.