Deep learning of representations and its application to computer vision

L’objectif de cette thèse par articles est de présenter modestement quelques étapes du parcours qui mènera (on espère) à une solution générale du problème de l’intelligence artificielle. Cette thèse contient quatre articles qui présentent chacun une différente nouvelle méthode d’inférence perceptive en utilisant l’apprentissage machine et, plus particulièrement, les réseaux neuronaux profonds. Chacun de ces documents met en évidence l’utilité de sa méthode proposée dans le cadre d’une tâche de vision par ordinateur. Ces méthodes sont applicables dans un contexte plus général, et dans certains cas elles on tété appliquées ailleurs, mais ceci ne sera pas abordé dans le contexte de cette de thèse. Dans le premier article, nous présentons deux nouveaux algorithmes d’inférence variationelle pour le modèle génératif d’images appelé codage parcimonieux “spike- and-slab” (CPSS). Ces méthodes d’inférence plus rapides nous permettent d’utiliser des modèles CPSS de tailles beaucoup plus grandes qu’auparavant. Nous démontrons qu’elles sont meilleures pour extraire des détecteur de caractéristiques quand très peu d’exemples étiquetés sont disponibles pour l’entraînement. Partant d’un modèle CPSS, nous construisons ensuite une architecture profonde, la machine de Boltzmann profonde partiellement dirigée (MBP-PD). Ce modèle a été conçu de manière à simplifier d’entraînement des machines de Boltzmann profondes qui nécessitent normalement une phase de pré-entraînement glouton pour chaque couche. Ce problème est réglé dans une certaine mesure, mais le coût d’inférence dans le nouveau modèle est relativement trop élevé pour permettre de l’utiliser de manière pratique. Dans le deuxième article, nous revenons au problème d’entraînement joint de machines de Boltzmann profondes. Cette fois, au lieu de changer de famille de modèles, nous introduisons un nouveau critère d’entraînement qui donne naissance aux machines de Boltzmann profondes à multiples prédictions (MBP-MP). Les MBP-MP sont entraînables en une seule étape et ont un meilleur taux de succès en classification que les MBP classiques. Elles s’entraînent aussi avec des méthodes variationelles standard au lieu de nécessiter un classificateur discriminant pour obtenir un bon taux de succès en classification. Par contre, un des inconvénients de tels modèles est leur incapacité de générer deséchantillons, mais ceci n’est pas trop grave puisque la performance de classification des machines de Boltzmann profondes n’est plus une priorité étant donné les dernières avancées en apprentissage supervisé. Malgré cela, les MBP-MP demeurent intéressantes parce qu’elles sont capable d’accomplir certaines tâches que des modèles purement supervisés ne peuvent pas faire, telles que celle de classifier des données incomplètes ou encore celle de combler intelligemment l’information manquante dans ces données incomplètes. Le travail présenté dans cette thèse s’est déroulé au milieu d’une période de transformations importantes du domaine de l’apprentissage à réseaux neuronaux profonds qui a été déclenchée par la découverte de l’algorithme de “dropout” par Geoffrey Hinton. Dropout rend possible un entraînement purement supervisé d’architectures de propagation unidirectionnel sans être exposé au danger de sur- entraînement. Le troisième article présenté dans cette thèse introduit une nouvelle fonction d’activation spécialement con ̧cue pour aller avec l’algorithme de Dropout. Cette fonction d’activation, appelée maxout, permet l’utilisation de aggrégation multi-canal dans un contexte d’apprentissage purement supervisé. Nous démontrons comment plusieurs tâches de reconnaissance d’objets sont mieux accomplies par l’utilisation de maxout. Pour terminer, sont présentons un vrai cas d’utilisation dans l’industrie pour la transcription d’adresses de maisons à plusieurs chiffres. En combinant maxout avec une nouvelle sorte de couche de sortie pour des réseaux neuronaux de convolution, nous démontrons qu’il est possible d’atteindre un taux de succès comparable à celui des humains sur un ensemble de données coriace constitué de photos prises par les voitures de Google. Ce système a été déployé avec succès chez Google pour lire environ cent million d’adresses de maisons.

Estado situacional de los modelos basados en agentes y su impacto en la investigación organizacional

En un mundo hiperconectado, dinámico y cargado de incertidumbre como el actual, los métodos y modelos analíticos convencionales están mostrando sus limitaciones. Las organizaciones requieren, por tanto, herramientas útiles que empleen tecnología de información y modelos de simulación computacional como mecanismos para la toma de decisiones y la resolución de problemas. Una de las más recientes, potentes y prometedoras es el modelamiento y la simulación basados en agentes (MSBA). Muchas organizaciones, incluidas empresas consultoras, emplean esta técnica para comprender fenómenos, hacer evaluación de estrategias y resolver problemas de diversa índole. Pese a ello, no existe (hasta donde conocemos) un estado situacional acerca del MSBA y su aplicación a la investigación organizacional. Cabe anotar, además, que por su novedad no es un tema suficientemente difundido y trabajado en Latinoamérica. En consecuencia, este proyecto pretende elaborar un estado situacional sobre el MSBA y su impacto sobre la investigación organizacional.

Didactics–promoting collaborative work and personal learning environment. Critical, creative and autonomous roles in a competitive society

School has evolved from a place where knowledge is provided to a place where learners are helped to develop their professional and social skills. Consequently, education must evolve through big challenges in order to face the changes of society in the XXIst century

Collaborative eLearning Assistant Network: caring agents are conscious agents

Would a research assistant - who can search for ideas related to those you are working on, network with others (but only share the things you have chosen to share), doesn’t need coffee and who might even, one day, appear to be conscious - help you get your work done? Would it help your students learn? There is a body of work showing that digital learning assistants can be a benefit to learners. It has been suggested that adaptive, caring, agents are more beneficial. Would a conscious agent be more caring, more adaptive, and better able to deal with changes in its learning partner’s life? Allow the system to try to dynamically model the user, so that it can make predictions about what is needed next, and how effective a particular intervention will be. Now, given that the system is essentially doing the same things as the user, why don’t we design the system so that it can try to model itself in the same way? This should mimic a primitive self-awareness. People develop their personalities, their identities, through interacting with others. It takes years for a human to develop a full sense of self. Nobody should expect a prototypical conscious computer system to be able to develop any faster than that. How can we provide a computer system with enough social contact to enable it to learn about itself and others? We can make it part of a network. Not just chatting with other computers about computer ‘stuff’, but involved in real human activity. Exposed to ‘raw meaning’ – the developing folksonomies coming out of the learning activities of humans, whether they are traditional students or lifelong learners (a term which should encompass everyone). Humans have complex psyches, comprised of multiple strands of identity which reflect as different roles in the communities of which they are part – so why not design our system the same way? With multiple internal modes of operation, each capable of being reflected onto the outside world in the form of roles – as a mentor, a research assistant, maybe even as a friend. But in order to be able to work with a human for long enough to be able to have a chance of developing the sort of rich behaviours we associate with people, the system needs to be able to function in a practical and helpful role. Unfortunately, it is unlikely to get a free ride from many people (other than its developer!) – so it needs to be able to perform a useful role, and do so securely, respecting the privacy of its partner. Can we create a system which learns to be more human whilst helping people learn?

Learning how older adults undertake computer tasks [poster]

This paper describes a study that was conducted to learn more about how older adults use the tools in a GUI to undertake tasks in Windows applications. The objective was to gain insight into what people did and what they found most difficult. File and folder manipulation, and some aspects of formatting presented difficulties, and these were thought to be related to a lack of understanding of the task model, the correct interpretation of the visual cues presented by the interface, and the recall and translation of the task model into a suitable sequence of actions.

Collaborative grid infrastructure for molecular simulations: The eMinerals minigrid as a prototype integrated computer and data grid

This paper describes a prototype grid infrastructure, called the eMinerals minigrid, for molecular simulation scientists. which is based on an integration of shared compute and data resources. We describe the key components, namely the use of Condor pools, Linux/Unix clusters with PBS and IBM's LoadLeveller job handling tools, the use of Globus for security handling, the use of Condor-G tools for wrapping globus job submit commands, Condor's DAGman tool for handling workflow, the Storage Resource Broker for handling data, and the CCLRC dataportal and associated tools for both archiving data with metadata and making data available to other workers.

Ontology supported competency system

Traditionally representation of competencies has been very difficult using computer-based techniques. This paper introduces competencies, how they are represented, and the related concept of competency frameworks and the difficulties in using traditional ontology techniques to formalise them. A “vaguely” formalised framework has been developed within the EU project TRACE and is presented. The framework can be used to represent different competencies and competency frameworks. Through a case study using an example from the IT sector, it is shown how these can be used by individuals and organisations to specify their individual competency needs. Furthermore it is described how these representations are used for comparisons between different specifications applying ontologies and ontology toolsets. The end result is a comparison that is not binary, but tertiary, providing “definite matches”, possible / partial matches, and “no matches” using a “traffic light” analogy.

Computer tools to aid the learning and design steps for photovoltaic systems

This paper presents novel simulation tools to assist the lecturers about learning processes on renewable energy sources, considering photovoltaic (PV) systems. The PV behavior, functionality and its interaction with power electronic converters are investigated in the simulation tools. The main PV output characteristics, I (current) versus V (voltage) and P (power) versus V (voltage), were implemented in the tools, in order to aid the users for the design steps. In order to verify the effectiveness of the developed tools the simulation results were compared with Matlab. Finally, a prototype was implemented with the purpose to compare the experimental results with the results from the proposed tools, validating its operational feasibility. © 2011 IEEE.

Building Leadership: The Knowledge of Principals in Creating Collaborative Communities of Professional Learning

Research literature is replete with the importance of collaboration in schools, the lack of its implementation, the centrality of the role of the principal, and the existence of a gap between knowledge and practice--or a "Knowing-Doing Gap." In other words, there is a set of knowledge that principals must know in order to create a collaborative workplace environment for teachers. This study sought to describe what high school principals know about creating such a culture of collaboration. The researcher combed journal articles, studies and professional literature in order to identify what principals must know in order to create a culture of collaboration. The result was ten elements of principal knowledge: Staff involvement in important decisions, Charismatic leadership not being necessary for success, Effective elements of teacher teams, Administrator‘s modeling professional learning, The allocation of resources, Staff meetings focused on student learning, Elements of continuous improvement, and Principles of Adult Learning, Student Learning and Change. From these ten elements, the researcher developed a web-based survey intended to measure nine of those elements (Charismatic leadership was excluded). Principals of accredited high schools in the state of Nebraska were invited to participate in this survey, as high schools are well-known for the isolation that teachers experience--particularly as a result of departmentalization. The results indicate that principals have knowledge of eight of the nine measured elements. The one that they lacked an understanding of was Principles of Student Learning. Given these two findings of what principals do and do not know, the researcher recommends that professional organizations, intermediate service agencies and district-level support staff engage in systematic and systemic initiatives to increase the knowledge of principals in the element of lacking knowledge. Further, given that eight of the nine elements are understood by principals, it would be wise to examine reasons for the implementation gap (Knowing-Doing Gap) and how to overcome it.

Using Machine Learning Classifiers to Assist Healthcare-Related Decisions: Classification of Electronic Patient Records

Surveillance Levels (SLs) are categories for medical patients (used in Brazil) that represent different types of medical recommendations. SLs are defined according to risk factors and the medical and developmental history of patients. Each SL is associated with specific educational and clinical measures. The objective of the present paper was to verify computer-aided, automatic assignment of SLs. The present paper proposes a computer-aided approach for automatic recommendation of SLs. The approach is based on the classification of information from patient electronic records. For this purpose, a software architecture composed of three layers was developed. The architecture is formed by a classification layer that includes a linguistic module and machine learning classification modules. The classification layer allows for the use of different classification methods, including the use of preprocessed, normalized language data drawn from the linguistic module. We report the verification and validation of the software architecture in a Brazilian pediatric healthcare institution. The results indicate that selection of attributes can have a great effect on the performance of the system. Nonetheless, our automatic recommendation of surveillance level can still benefit from improvements in processing procedures when the linguistic module is applied prior to classification. Results from our efforts can be applied to different types of medical systems. The results of systems supported by the framework presented in this paper may be used by healthcare and governmental institutions to improve healthcare services in terms of establishing preventive measures and alerting authorities about the possibility of an epidemic.

Robotics learning based on collaborative projects

[EN]Nowadays companies demand graduates able to work in multidisciplinary and collaborative projects. Hence, new educational methods are needed in order to support a more advanced society, and progress towards a higher quality of life and sustainability. The University of the Basque Country belongs to the European Higher Education Area, which was created as a result of the Bologna process to ensure the connection and quality of European national educational systems. In this framework, this paper proposes an innovative teaching methodology developed for the "Robotics" subject course that belongs to the syllabus of the B.Sc. degree in Industrial Electronics and Automation Engineering. We present an innovative methodology for Robotics learning based on collaborative projects, aimed at responding to the demands of a multidisciplinary and multilingual society.