Streamlining the medication process improves safety in the intensive care unit.

BACKGROUND: Multiple interventions were made to optimize the medication process in our intensive care unit (ICU). 1 Transcriptions from the medical order form to the administration plan were eliminated by merging both into a single document; 2 the new form was built in a logical sequence and was highly structured to promote completeness and standardization of information; 3 frequently used drug names, approved units, and fixed routes were pre-printed; 4 physicians and nurses were trained with regard to the correct use of the new form. This study was aimed at evaluating the impact of these interventions on clinically significant types of medication errors. METHODS: Eight types of medication errors were measured by a prospective chart review before and after the interventions in the ICU of a public tertiary care hospital. We used an interrupted time-series design to control the secular trends. RESULTS: Over 85 days, 9298 lines of drug prescription and/or administration to 294 patients, corresponding to 754 patient-days were collected and analysed for the three series before and three series following the intervention. Global error rate decreased from 4.95 to 2.14% (-56.8%, P < 0.001). CONCLUSIONS: The safety of the medication process in our ICU was improved by simple and inexpensive interventions. In addition to the optimization of the prescription writing process, the documentation of intravenous preparation, and the scheduling of administration, the elimination of the transcription in combination with the training of users contributed to reducing errors and carried an interesting potential to increase safety.

Teaching Principled negotiation skills to parents and their children : the impact of parental involvement /

This study explored the impact of training parents and children concurrently in principled negotiation skills for the purpose of developing negotiation skills and problem solving abilities in children. A second experimental group was utilized to determine the viability of negotiation skills training of junior elementary students for the purpose of improving problem solving and conflict resolving abilities. The student population in each experimental group was trained using The Program for Young Negotiators (Curhan, 1996). A control group was also established using the remaining grade four and five students attending the participating school. These students did not receive training as part of this study. Student group distribution was as follows: Experimental group 1 (students with parent participant) consisted of 10 (5 grade five and 5 grade 4 students), Experimental group 2 students without parent participant) consisted of 48 (20 grade 4 and 28 grade 5 students), and the Control group 3 (55 grade 4 and 5 students). The impact of training was measured using the Five Factor Negotiation Scale developed for use with the Program for Young Negotiators (Curhan, 1996). This measure was employed as a pre- and post-test questionnaire to the total student population, (113 students) to determine levels of ability in each of the key elements of negotiation, personal initiative, collaboration, communication, conflict based perspective taking, and conflict resolution approach (Nakkula & Nikitopoulos, unpublished). This measure has a coefficient alpha of .75 which is acceptable for this type of affective instrument. As well, open ended ability questions designed to measure ability, knowledge, and behaviour as they relate to negotiation skill application were given to the total student population, (113 students). Finally, journals were maintained by the students in both experimental groups, and informal feedback discussions were held with students and parents participating in the study.The intent of using both qualitative and quantitative measures was to provide an overall perspective of student abilities as they related to principled negotiation skills. While the quantitative measures were from the student perspective, more qualitative information was sought from parents and teachers through informal interviews, discussions, and use of confidential feedback cards. For analysis purposes, the ability questions were randomly selected for Experimental group 2 and Control group 3 in an effort to balance the groups more equitably with Experimental group 1. The findings of this study indicate that students of the junior elementary school age can be taught how to perceive conflict in a more constructive way. However, they are not as likely to use their skills when the conflict is with a sibling as they are with a peer, a teacher, or a parent. While no statistically significant differences between mean scores for Experimental groups 1 and 2 exist some subtle differences are noted. Overall, increases in mean scores for grade 4 students exceeded the increases for grade 5 students within Experimental group 1 . The implication being that younger students benefit more from having a parent trained in principled negoUation skills than older students. The skill level of a parent in principled negotiation can not be underesUmated. Without a consistent and effective role model the likelihood of developing student skill level to a point of automaticity is greatly reduced. Enough so that perhaps the emphasis should be placed on training parents more so than the students.

An after-school literacy program : investigating the experiences of students with literacy difficulties, their volunteer tutors, and the tutors' transition into the teaching profession

This research responds to a pervasive call for our educational institutions to provide students with literacy skills, and teachers with the instructional supports necessary to facilitate this skill acquisition. Questions were posed to gain information concerning the efficacy ofteaching literacy strategies to students with learning difficulties, the impact of this training on their volunteer tutors, and the influence of this experience on these tutors' ensuing instructional practice as teacher candidates in a preservice education program. Study #1 compared a nontreatment group of students with literacy difficulties who participated in the program and found that program participants were superior at reading letter patterns and at comprehending the elements of story grammar. Concurrently, the second study explored the experiences of 19 volunteer tutors and uncovered that they acquired instructional skills as they established a knowledge base in teaching reading and writing, and they affirmed personal goals to become future teachers. Study #3 tracked 6 volunteer tutors into their pre-service year and identified their constructions, and beliefs about literacy instruction. These teacher candidates discussed how they had intended to teach reading and writing strategies based on their position that effective teaching ofthese skills in the primary grades is integral to academic success. The teacher candidates emphasized the need to build rapport with students, and the need to exercise flexibility in lesson plan delivery while including activities to meet emotional and developmental requirements of students. The teacher candidates entered their pre-service education with an initial cognition set based on the limited teaching context of tutoring. This foundational ii perception represented their prior knowledge of literacy instruction, a perception that appeared untenable once they were immersed in a regular instructional setting. This disparity provoked some of the teacher candidates to denounce their teacher mentors for not consistently employing literacy strategies and individualized instruction. This critical perspective could have been a demonstration of cognitive dissonance. In the end, when the teacher candidates began to look toward the future and how they would manage the demands of an inclusive classroom, they recognized the differences in the contexts. With an appreciation for the need for balance between prior and present knowledge, the teacher candidates remained committed to implementing their tutoring strategies in future teaching positions. This document highlights the need for teacher candidates with instructional experience prior to teacher education, to engage in cognitive negotiations to assimilate newly acquired pedagogies into existing pedagogies.

Novice teachers' perceptions of the extent to which the Brock University teacher education program focused on methods of promoting responsibility in students

The purpose of this study was to determine novice t~ache~s' perceptions of th~ extent to which the Brock University teacher education program focused on strategies for promoting responsibility in students. Individual interviews were conducted with ten randomly selected teachers who were graduates of this teacher education program between the years of 1989 and 1992, and a follow-up group discussion activity, with the same teachers, was also held. Findings revealed that the topic of personal responsibility was discussed within various components of the program, including counselling group sessions, but that these discussions were often brief, indirect and inconsistent. Some of the strategies which the teachers used in their own classrooms to promote responsibility in students were ones which they had acquired from those counselling group °sessions or from associate teachers. Various strategies included: setting ~lear expectations of students with positive and negative consequences for behaviour (e.g., material rewards and detentions, respectively), cemmunic?ting'with other teachers an~ parents, and -. suspending students from school. A teacher's choice of any particular strategy seemed to be affected by his or her personality, teaching sUbject and region of employment, as well as certain aspects of the teacher education program. It was concluded that many of the teachers appeared to be controlling rude and vio~ent- behaviour, as opposed to promoting responsible behaviour. Recommendations were made for the pre-service program, as well as induction and inservice programs, to increase teacher preparedness for promoting responsible student behaviour. One of these recommendations addressed the need to help teachers learn how to effectively communicate with their students.

The efficacy of cognitive retraining of attentional deficits for a client who has schizophrenia

The purpose of this study was to assess the efficacy of the Process Specific Approach to cognitive rehabilitation for a client with schizophrenia who has attentional deficits. The study was a single case experimental design which followed a variation of the multiple baseline approach. Prior to training of the attentional deficit, multiple baseline assessments were completed. These included an ov:erview of the sUbject's information processing ability, random measures of attention and a genera.l level of functioning in living, learning and working environments. During the re-training, attention tests were administered at the completion of each attention component. A general functional evaluation through interviews and a measure of information processing ability were. completed after the re-training was concluded. The results of the study demonstrate a significant i'mprovement in attention and memory measures. Qualitative data indicate si·gni.ficant others observed improvements in performance in r livi'ng, learning and working environments. The results suggest this approach to cognitive rehabilitation was effective with this subject and further research to establish generalizability is recommended.

A study of the role learning and socialization during the transition from teacher to elementary school administrator

This thesis examined the role transition from an elementary teacher to an elementary principal. In particular, the training and socialization process of becoming an elementary principal was explored through the study of the hierarchical and political structure of a southern Ontario school board, and how this influenced the learning experiences of new elementary principals. A qualitative methodology, with a grounded theory design, was employed to investigate this process through interviews with 10 participants to examine their experiences and role learning occurs during their development. Specifically, participants perspective shifts, developmental experiences, understanding of group culture, and expansion of a board profile were highlighted in the data. One of the compelling results of the study was the degree to which principals of aspiring administrators influence the socialization of their subordinates. The beliefs and practices of the school principal determine the socialization orientation that teachers and vice-principals will experience during role learning. The results of this study also imply that role orientation needs to be understood as a continuum between custodial and innovative role assumption. Varying degrees of custodianship or innovation depended on the context of the administrative placement and the personal attributes of administrative candidates. Principals who are willing to share responsibilities, who are good communicators, and who wish to develop a collaborative relationship with their viceprincipals are the individuals the participants in this study described as making the best mentors.

Show me, help me, let me : supporting teachers' changing conceptions of reading assessment and reading instruction

This qualitative inquiry used case study methodology to explore the change processes of 3 primary-grade teachers throughout their participation in 7 -month professional learning initiative focused on reading assessment and instruction. Participants took part in semimonthly inquiry-based professional learning community sessions, as well as concurrent individualized classroom-based literacy coaching. Each participant's experiences were first analyzed as a single case study, followed by cross-case analyses. While their patterns of professional growth differed, findings documented how all participants altered their understandings of the roles and relevancy of individual components of reading instruction (e.g., comprehension, decoding) and instructional approaches to scaffold students' growth (e.g., levelled text, strategy instruction), and experienced some form of conceptual change. Factors identified as affecting their change processes included; motivation, professional knowledge, professional beliefs (self-efficacy and theoretical orientation), resources (e.g., time, support), differentiated professional learning with associated goal-setting, and uncontrollable influences, with the affect of each factor compounded by interaction with the others. Comparison of participants' experiences to the Cognitive-Affective Model of Conceptual Change (CAMCC) and the Interconnected Model of Teacher Professional Growth (IMTPG) demonstrated the applicability of using both conceptual models, with the IMTPG providing macrolevel insights over time and the CAMCC microlevel insights at each change intervaL Recommendations include the provision of differentiated teacher professional learning opportunities, as well as research documenting the effects of teacher mentorship programs and the professional growth of teacher educators. ii

The expression and development of teachers' capacities within two learning communities : a participant-observer case study

The learning community model has been an integral component of teacher development in Ontarian schools and beyond. This research was conducted to understand how teachers' personal capacity and professional, interpersonal, and organizational competencies are developed and expressed within this context. Nineteen elementary teachers and administrators participated in the study from November through January 2007. A qualitative case study methodology was used to investigate the role ofteachers' capacities and competencies in learning communities. Combined data sources from semistructured interviews, research journals, and document review were used to gather data about teachers' capacities and competencies. The study included 3 phases of analysis. In the final phase the analysis provided 3 qualities of the teachers at Jude and Mountain Schools (pseudonyms): identification as professionals, investment in others, and institutional affiliation that may explain how they differed from other educators. The data revealed these three themes, which provided an understanding of educators at Jude and Mountain Schools as dedicated professionals pushing practices to contribute to school life and address student learning needs, and as teachers who reflected on practices to continue expanding their skills. Teachers were heavily invested in creating a caring culture and in students' and team members' learning. Educators actively participated in solving problems and coplanning throughout the school levels and beyond, assumed collective responsibility for all pupils, and focused on generating school-wide consistent practices. These qualities and action patterns revealed teachers who invested time and effort in their colleagues, who committed to develop as professionals, and who affiliated closely with every aspect of school living.

Teachers' professional learning within communities of mathematics-for-teaching practice

Ontario bansho is an emergent mathematics instructional strategy used by teachers working within communities of practice that has been deemed to have a transformational effect on teachers' professional learning of mathematics. This study sought to answer the following question: How does teachers' implementation of Ontario bansho within their communities of practice inform their professional learning process concerning mathematics-for-teaching? Two other key questions also guided the study: What processes support teachers' professional learning of content-for-teaching? What conditions support teachers' professional learning of content-for-teaching? The study followed an interpretive phenomenological approach to collect data using a purposive sampling of teachers as participants. The researcher conducted interviews and followed an interpretive approach to data analysis to investigate how teachers construct meaning and create interpretations through their social interactions. The study developed a model of professional learning made up of 3 processes, informing with resources, engaging with students, and visualizing and schematizing in which the participants engaged and 2 conditions, ownership and community that supported the 3 processes. The 3 processes occur in ways that are complex, recursive, nonpredictable, and contextual. This model provides a framework for facilitators and leaders to plan for effective, content-relevant professional learning by placing teachers, students, and their learning at the heart of professional learning.

Breaking the silence : exploring the workplace experiences of six women with learning disabilities

In the current economic climate, employees are expected to upgrade their skills in order to remain productive and competitive in the workplace, and many women with learning disabilities! may feel doubly challenged when dealing with such expectations. Although the number of people with reported learning disabilities who enter the workforce is expected to increase, a dearth of research focuses on work-related experiences of women with learning disabilities; consequently, employers and educators often are unaware ofthe obstacles and demands facing such individuals. This qualitative narrative study sheds light on the work experiences of women with diagnosed or suspected learning disabilities. The study used semistructured interviews to explore their perspectives and reflections on learnlng in order to: (a) raise awareness of the needs of women with learning disabilities, (b) enhance their opportunities to learn in the workplace, and (c) draw attention to the need for improvement of inclusiveness in the workplace, especially for hidden disabilities. Study findings reveal that participants' learning was influenced by work relationships, the learning environments, self-determination, and taking personal responsibility. Moreover, the main accommodation requested was to have supportive and understanding work relationships and environments. Recommendations are made for future research and workplace improvements, most notably that no employees should be left behind through an employee-centered approach.

A district school board's induction program for new administrators: a study of the effectiveness of the mentoring component

Please consult the paper edition of this thesis to read. It is available on the 5th Floor of the Library at Call Number: Z 9999 E38 K535 2008

Training deep convolutional architectures for vision

Les tâches de vision artificielle telles que la reconnaissance d’objets demeurent irrésolues à ce jour. Les algorithmes d’apprentissage tels que les Réseaux de Neurones Artificiels (RNA), représentent une approche prometteuse permettant d’apprendre des caractéristiques utiles pour ces tâches. Ce processus d’optimisation est néanmoins difficile. Les réseaux profonds à base de Machine de Boltzmann Restreintes (RBM) ont récemment été proposés afin de guider l’extraction de représentations intermédiaires, grâce à un algorithme d’apprentissage non-supervisé. Ce mémoire présente, par l’entremise de trois articles, des contributions à ce domaine de recherche. Le premier article traite de la RBM convolutionelle. L’usage de champs réceptifs locaux ainsi que le regroupement d’unités cachées en couches partageant les même paramètres, réduit considérablement le nombre de paramètres à apprendre et engendre des détecteurs de caractéristiques locaux et équivariant aux translations. Ceci mène à des modèles ayant une meilleure vraisemblance, comparativement aux RBMs entraînées sur des segments d’images. Le deuxième article est motivé par des découvertes récentes en neurosciences. Il analyse l’impact d’unités quadratiques sur des tâches de classification visuelles, ainsi que celui d’une nouvelle fonction d’activation. Nous observons que les RNAs à base d’unités quadratiques utilisant la fonction softsign, donnent de meilleures performances de généralisation. Le dernière article quand à lui, offre une vision critique des algorithmes populaires d’entraînement de RBMs. Nous montrons que l’algorithme de Divergence Contrastive (CD) et la CD Persistente ne sont pas robustes : tous deux nécessitent une surface d’énergie relativement plate afin que leur chaîne négative puisse mixer. La PCD à "poids rapides" contourne ce problème en perturbant légèrement le modèle, cependant, ceci génère des échantillons bruités. L’usage de chaînes tempérées dans la phase négative est une façon robuste d’adresser ces problèmes et mène à de meilleurs modèles génératifs.

Measuring and evaluating quality of care in referral maternities in Mali and Senegal in the context of overlapping interventions

Dans cette thèse, nous décrivons les résultats d’un projet de recherche visant à mesurer et évaluer la qualité des soins obstétricaux des hôpitaux de référence au Mali et au Sénégal. Dans ces pays, la mortalité maternelle hospitalière est élevée et est liée en partie à la pratique médicale inadéquate. Cette recherche a été réalisée dans le cadre de l’étude QUARITE, un essai randomisé en grappe évaluant l’efficacité du programme GESTA International visant à réduire la mortalité maternelle hospitalière. GESTA a été mis en œuvre entre 2008 et 2010 et consistait en la formation des professionnels de santé et en la revue des cas de décès maternels. En parallèle de QUARITE, les programmes de prévention de la transmission du VIH de la mère à l’enfant (PTME) ont été mis à l’échelle à travers les pays. Ces derniers ayant également la capacité d’augmenter la qualité des soins obstétricaux, nous avons donc évalué les effets des deux programmes (GESTA et PTME) sur la qualité des soins. Dans un premier temps, à l’aide d’une recension des écrits nous avons évalué la capacité d’un audit clinique basé sur des critères à mesurer la qualité des soins obstétricaux. Cet audit vérifiait si l’offre des soins avait respecté les critères cliniques définissant la meilleure prise en charge selon l’évidence scientifique et l’avis des experts. Nous avons démontré que cet outil est largement utilisé dans les pays à faibles et moyens revenus, malgré le peu d’évidence sur sa validité (article 1). Dans un deuxième temps, nous avons développé un audit clinique basé sur des critères qui s’applique au contexte ouest-africain et qui a été approuvé par des experts-obstétriciens nationaux et internationaux. À partir des dossiers obstétricaux, les actes médicaux posés pendant le travail et l’accouchement ont été évalués à l‘aide de cet instrument. La qualité des soins a été estimée sous forme de pourcentage de critères atteints. Appliqué dans différents contextes et par différents auditeurs, nous avons démontré que notre instrument est fiable et valide (article 3). Néanmoins, l’expérience de l’audit nous a amenés à nous questionner sur le mauvais remplissage des dossiers médicaux et ses conséquences sur la qualité des soins (article 2). Dans un troisième temps, l’outil a été appliqué à large échelle pour évaluer les effets de l’intervention GESTA (article 4). Nous avons mené une révision de plus de 800 dossiers obstétricaux dans 32 hôpitaux de référence (16 bénéficiaires de l’intervention et 16 non-bénéficiaires). Grâce à cet audit clinique, nous avons démontré que le programme GESTA contribue à l’amélioration de la qualité des soins, spécifiquement l’examen clinique lors de l’admission et le suivi après l’accouchement. Dernièrement, nous avons utilisé cet instrument afin d’évaluer les effets des programmes de PTME sur la qualité des soins obstétricaux (article 5). Notre travail a documenté que seulement certaines composantes du programme de PTME améliorent la qualité des soins telles que la formation des professionnels et les services complémentaires en nutrition. En conclusion, cette recherche a identifié plusieurs pistes d’intervention pour améliorer la qualité des soins obstétricaux en Afrique de l’Ouest.

Portrait sociodémographique, anamnestique et psychopathologique des enfants placés en Centre jeunesse et à qui l'on prescrit des médicaments psychotropes.

Cette thèse a pour but de mieux comprendre le recours aux médicaments psychotropes chez les enfants hébergés en Centre jeunesse, une population qui compte parmi les plus médicalisées. Afin d’y parvenir, les caractéristiques sociodémographiques, anamnestiques et psychopathologiques des enfants placés qui reçoivent des psychopharmacoprescriptions ont été analysées. L’échantillon se compose de 101 enfants de 6 à 12 ans placés en foyer de groupe ou centre de réadaptation, dont 71 recevant des médicaments psychotropes. Les données ont été obtenues par le biais de questionnaires remplis par les éducateurs, par une entrevue semi-structurée réalisée auprès des enfants et une analyse du dossier. Sur le plan de la structure, cet ouvrage comprend une introduction, quatre articles et une conclusion. Le premier article comporte une recension des écrits sur l’usage de la psychopharmacothérapie chez les jeunes placés dans les services de la protection de la jeunesse. Il a été constaté que les taux de prescriptions peuvent varier entre 13% et 77%, selon le type de placement et les régions à l’étude. La symptomatologie des enfants placés qui reçoivent des médicaments psychotropes est caractérisée par des problèmes extériorisés et des troubles psychotiques. Les corrélats du recours à la prescription concernent à la fois la sévérité du tableau clinique, mais aussi le type et l’instabilité du placement, l’âge et le sexe de l’enfant ainsi que la formation des intervenants. Enfin, les écrits recensés font état de l’influence des neurosciences dans les milieux médicaux sur les décisions de prescrire. Le deuxième article présente la prévalence des psychopharmacoprescriptions chez les enfants de 6 à 12 ans placés hors d’un milieu familial. Les résultats indiquent que 70,3% des jeunes reçoivent au moins une prescription, le plus souvent signées par des médecins spécialistes. La plupart se composent de psychostimulants et d’antipsychotiques atypiques, prescrits pour des troubles de l’attention avec hyperactivité. Le troisième article cherche à préciser les caractéristiques sociodémographiques et anamnestiques des enfants placés qui prennent des médicaments psychotropes. Les résultats indiquent que les sujets médicamentés et non médicamentés ont vécu des stresseurs psychosociaux similaires. Par contre, les enfants placés qui reçoivent une psychopharmacothérapie ont été retirés de leur milieu familial à un plus jeune âge. Le quatrième article consiste à cibler leur portrait psychopathologique et à connaître leur niveau de fonctionnement global. Il s’intéresse également aux connaissances et aux perceptions des éducateurs sur la psychopharmacothérapie. Les résultats révèlent que les enfants médicamentés ont plus souvent un diagnostic de trouble mental inscrit à leur dossier. Selon les éducateurs, ils présentent davantage de problèmes extériorisés et intériorisés. À partir de données autorévélées, aucune distinction ne peut être établie entre les sujets, puisque les uns et les autres rapportent un niveau comparable de symptômes et de signes diagnostiques. Enfin, les symptômes de stress post-traumatique et l’opinion favorable des éducateurs sur la psychopharmacothérapie constituent des prédicteurs significatifs de la probabilité de recourir à un traitement médicamenteux. En conclusion, l’apport des résultats de cette recherche est analysé à la lumière des études antérieures. Les retombées cliniques sont discutées et des pistes de recherche futures sont suggérées. .

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.