This thesis presents a deep learning tool able to identify ice in radar images fromthe sea-ice environment of the Twhaites glacier outlet. The project is motivatedby the threatening situation of the Thwaites glacier that has been increasingits mass loss rate during the last decade. This is of concern considering thelarge mass of ice held by the glacier, that in case of melting, could increasethe mean sea level by more than +65 cm [1]. The algorithm generated alongthis work is intended to help in the generation of navigation charts and identificationof icebergs in future stages of the project, outside of the scope of this thesis.The data used for this task are ICEYE’s X-band radar images from the Thwaitessea-ice environment, the target area to be studied. The corresponding groundtruth for each of the samples has been manually generated identifying the iceand icebergs present in each image. Additional data processing includes tiling,to increment the number of samples, and augmentation, done by horizontal andvertical flips of a random number of tiles.The proposed tool performs semantic segmentation on radar images classifyingthe class "Ice". It is developed by a deep learning Convolutional Neural Network(CNN) model, trained with prepared ICEYE’s radar images. The model reachesvalues of F1 metric higher than 89% in the images of the target area (Thwaitessea-ice environment) and is able to generalize to different regions of Antarctica,reaching values of F1 = 80 %. A potential alternative version of the algorithm isproposed and discussed. This alternative score F1 values higher than F1 > 95 %for images of the target environment and F1 = 87 % for the image of the differentregion. However, it must not be confirmed as the final algorithm due to the needfor further verification.

Severe weather events like hurricanes and tornadoes pose major risks globally, underscoring the critical need for accurate forecasts to mitigate impacts. While advanced computational capabilities and climate models have improved predictions, lack of high-resolution initial conditions still limits forecast accuracy. The Atlantic's "Hurricane Alley" region sees most storms arise, thus needing robust in-situ ocean data plus atmospheric profiles to enable precise hurricane tracking and intensity forecasts. Examining satellite datasets reveals radio occultation (RO) provides the most accurate 5-25 km altitude atmospheric measurements. However, below 5 km accuracy remains insufficient over oceans versus land areas. Some recent benchmark study e.g. Patil Iiyama (2022), and Wei Guan (2022) in their work proposed the use of deep learning models for sea surface temperature (SST) prediction in the Tohoku region with very low errors ranging from 0.35°C to 0.75°C and the root-mean-square error increases from 0.27°C to 0.53°C over the over the China seas respectively. The approach we have developed remains unparalleled in its domain as of this date. This research is divided into two parts and aims to develop a data driven satellite-informed machine learning system to combine high-quality but sparse in-situ ocean data with more readily available low-quality satellite data. In the first part of the work, a novel data-driven satellite-informed machine learning algorithm was implemented that combines High-Quality/Low-Coverage in-situ point ocean data (e.g. ARGO Floats) and Low-Quality/High-Coverage Satellite ocean Data (e.g. HYCOM, MODIS-Aqua, G-COM) and generated high resolution data with a RMSE of 0.58◦C over the Atlantic Ocean.The second part of the work a novel GNN algorithm was implemented on the Gulf of Mexico and showed it can successfully capture the complex interactions between the ocean and mimic the path of a ARGO floats with a RMSE of 1.40◦C. / Doctor of Philosophy / Severe storms like hurricanes and tornadoes are a major threat around the world. Accurate weather forecasts can help reduce their impacts. While climate models have improved predictions, lacking detailed initial conditions still limits forecast accuracy. The Atlantic's "Hurricane Alley" sees many storms form, needing good ocean and atmospheric data for precise hurricane tracking and strength forecasts. Studying satellite data shows radio occultation provides the most accurate 5-25 km high altitude measurements over oceans. But below 5 km accuracy remains insufficient versus over land. Recent research proposed using deep learning models for sea surface temperature prediction with low errors. Our approach remains unmatched in this area currently. This research has two parts. First, we developed a satellite-informed machine learning system combining limited high-quality ocean data with more available low-quality satellite data. This generated high resolution Atlantic Ocean data with an error of 0.58°C. Second, we implemented a new algorithm on the Gulf of Mexico, successfully modeling complex ocean interactions and hurricane paths with an error of 1.40°C. Overall, this research advances hurricane forecasting by combining different data sources through innovative machine learning techniques. More accurate predictions can help better prepare communities in hurricane-prone regions.

Dans ce mémoire, nous nous penchons sur les processus créatifs et techniques d’enregistrement sonore de type DIY (Do It Yourself) pratiqués dans mes expériences avec différents groupes rock, punk, hardcore et alternatif de Québec. Les styles musicaux varient grandement d’un projet à l’autre, mais trois aspects sont communs à chaque groupe : 1) Ils sont issus d’une scène «underground» de la Ville de Québec. 2) Ils ont une instrumentation rock (pour nous, une instrumentation rock est composée des instruments suivants : batterie, guitare électrique, guitare basse et voix). 3) Ils enregistrent de manière DIY jusqu’à un certain niveau. Avec les avancées technologiques, l’enregistrement DIY est une pratique de plus en plus courante chez les groupes de musique. À notre expérience, les artistes œuvrant dans des scènes se rapprochant de la culture punk et underground sont souvent restreints par des ressources et des budgets limités ou ont simplement un désir de liberté créative lorsqu’il est temps d’enregistrer leurs chansons. Dans la recherche de l’indépendance, ils vont souvent opter pour l’approche DIY lorsqu’il est temps d’enregistrer une chanson ou un album, ce qui donne souvent une esthétique punk aux pistes audio. Dans ce partage d’expériences, l’objectif est d’analyser les méthodes de travail qui contribuent à obtenir des résultats de qualité en enregistrement DIY et d’en tirer une méthode efficace qui pourrait servir au DIYer en enregistrement. Nous croyons qu’en prenant connaissance de nos résultats et de nos analyses de chaque projet, un(e) artiste désirant enregistrer un projet musical de façon DIY sera en mesure de faire des choix plus sensés et éclairés par rapport à sa propre situation et son propre projet d’enregistrement. Enfin, j’offre mes observations et conclusions au réalisateur DIY pour qu’il analyse bien sa réalité unique à lui-même et à son ou ses projets d’enregistrement afin qu’il bénéficie d’une courbe d’apprentissage rapide et conviviale. / In this master’s thesis, we address the issue of creative processes and recording techniques in different DIY (Do It Yourself) settings as implemented through my experience with different rock, punk, hardcore and alternative bands in Quebec City. Musical styles greatly vary from one project to another, but all of the music groups share three common aspects : 1) They come from the underground scene around Quebec City. 2) They have a rock-type instrumentation (for us, a rock-type instrumentation is made up of minimum two of the following instruments: drums, electric guitar, electric bass and vocal.) 3) They record in a DIY setting, up to a certain point. With the rise of new technologies, DIY recording is more and more common in music bands. In our experience, artists from scenes close to the punk and underground cultures are often limited by their budget and resources or simply have a desire for creative liberty when recording their music. Seeking independence, they will often go for the DIY approach when the time has come for them to record a song or an album, which often lends a punk aesthetic to their audio tracks. This masters thesis seeks to analyse working methods that contribute to obtaining good quality results in DIY recording and to find an efficient methodology that could help the self-taught DIYer in recording music. We believe that by reading about these projects, with the analysis and feedback, an artist who wants to work in a DIY setting will be able to make better decisions in regard to their own recording project and situation. I also offer advice for the DIY recording artist to analyse his own unique recording reality and the one of his different recording projects in order for him to benefit from a faster and favorable learning curve.

Applying machine learning to various applications has gained significant momentum in recent years. However, the increasing complexity of networks introduces challenges such as a larger memory footprint and decreased throughput. This thesis aims to address these challenges by exploring the use of 8-bit floating-point numbers for machine learning. The numerical accuracy was evaluated empirically by implementing software models of the arithmetic and running experiments on a neural network provided by MediaTek. While the initial findings revealed poor accuracy when performing computations solely with 8-bit floating-point arithmetic, a significant improvement could be achieved by using a higher-precision accumulator register. The hardware cost was evaluated using a synthesis tool by measuring the increase in silicon area and impact on clock frequency after four new vector instructions had been implemented. A large increase in area was measured for the functional blocks, but the hardware cost for interconnect and instruction decoding were negligible. A slight decrease in system clock frequency was observed, although marginally. Ideas that likely could improve the accuracy of inference calculations and decrease the hardware cost are proposed in the section for future work.

Bandgap determines the suitability of materials for device applications such as \cite{lyu2021predictive} light-emitting diodes (LED), solar cells, and photo-detectors. The accuracy of the bandgap predicted using standard LDA or GGA functional is underestimated by density functional theory (DFT) when compared to experimental values. However, DFT combined with Machine Learning (ML) allows computational screening of materials with better accuracy. The training data for the models is obtained from density functional theory calculations which consist of A, B, and X-site elemental properties. The feature importance procedure screens the relative important features among all input features considered in the study. CatBoost(CB) regression model, \cite{Catboost} is an open-source library for gradient boosting. It gives high-perfromance on decision tress as it is based on gradient boosting algorithm and is suitable for small data sets as reduces overfitting, is implemented that can accurately predict the bandgap of $A_2BX_6$ perovskite. Eleven ML techniques were implemented, and their predictions of energy bandgap were compared, such as gradient boosted, random forest, support vector, AdaBoost, linear, K-nearest neighbor, kernel ridge, and decision tree. As per the study, the best performance is achieved by CatBoost regression model.

Cancer may be treated with personalized medicine, meaning that specific patientsmight respond better to specific treatments instead of having a common treatment. TheReference Drug-based Neural Network (RefDNN) predicts whether a particular cancer cellline will resist a determined drug, but it fails to provide an explanations for this prediction.The thesis objective is to research on explainable machine learning methods to extractrule-based explanations from the RefDNN predictions and conclude on how confident wecan be about these explanations and whether they make sense from a biological point ofview. One of such explainable machine learning methods is Local Rule-based Explanation(LORE), which extracts rule-based explanations from any black box model using localdecision trees. In this thesis LORE is applied to explain the predictions of the RefDNNon a drug sensitivity dataset and three experiments are set up. First experiment tests theaccuracy and general performance of the extracted rule-based explanations. Second experimentstests the robustness of the rule-based explanations. Third experiments checks theglobal fidelity of the local decision trees used by LORE to mimic the RefDNN behaviour.Finally, one rule-based decision is explained from a biological point of view and conclusionsare made on the obtained results.

<p dir="ltr">The Air Traffic Management (ATM) system encompasses complex and safety-critical operations which are mainly managed by Air Traffic Controllers (ATCs) and pilots to ensure safety and efficiency. This air traffic operation becomes more complex and challenging as demands continue to increase. Indeed, the demand for air transport is expected to increase by an average of 4.3% annually over the next 20 years, and the projected number of flights is expected to reach around 90 million by 2040 [1]. This continuous growth of demands can lead to an excessive workload for both ATCs and pilots, thereby resulting in the degradation of the ATM system. To effectively respond to this problem, a lot of effort has been put into developing decision support tools. This dissertation explores and focuses on the development of algorithms for decision support tools in air traffic control, emphasizing specific desirable properties essential for tasks such as tracking the position of aircraft and monitoring air traffic. The primary focus of this dissertation is to combine a data-driven model and a physics-based model systematically, thereby addressing the limitations of previous works in trajectory prediction and anomaly detection. Through a literature review, important properties, including real-time applicability, interpretability, and feasibility, are identified and pursued for practical applications. These properties are integrated into the proposed algorithms which combine data-driven and physics-based models to address dynamic air traffic scenarios effectively. To meet the requirement of real-time applicability, the algorithms are designed to be computationally efficient and adaptable to continuously changing conditions, ensuring timely provision of immediate information and near-instantaneous responses to assist ATCs. Subsequently, interpretability allows controllers to understand the reasoning behind the algorithm’s predictions. This is facilitated by the use of attention mechanisms and explicit physics-based guidance, making the predictions more intuitive and understandable. In addition, anomaly detection algorithms provide human-readable decision boundaries for flight states for a clear understanding. Lastly, feasibility ensures that the algorithms generate realistic aircraft trajectory predictions based on current flight states and air traffic conditions. This is achieved by physics-guided machine learning which leverages both data-driven and physics-based approaches, accounting for the aircraft dynamics and uncertainties. Moreover, practical and operational considerations are integrated into algorithms for real-world applications. This includes developing anomaly detection models that are adaptable to dynamic trajectory patterns to address the complexities of flexible area navigation airspace. Additionally, to reduce the workload of ATCs, providing immediate advisories for anomaly resolution and arrival sequencing is targeted by learning from historical data. By considering these properties with practical considerations, the dissertation presents a suite of algorithms that can effectively support human operators for air traffic control. </p>

Cette thèse explore les liens entre le country-western produit au Québec entre 1942 et 1957 et la modernité populaire. À l’aide de l’analyse musicale et de l’histoire, ce travail de recherche tente de cerner la signification culturelle du genre au moment de son émergence. L’histoire du country-western débute au Québec avec le soldat Roland Lebrun, qui amorce en 1942 sa carrière sur disque. Il sera suivi de Paul Brunelle, de Marcel Martel et de Willie Lamothe, qui enregistrent chez l’une ou l’autre des deux compagnies généralistes établies à Montréal pendant les années 1940, Compo et RCA Victor. À mesure que le genre se structure, notamment grâce à la fondation de compagnies de disques spécialisées à partir de 1958, un discours sur l’authenticité du country-western se développe chez les artistes et leurs observateurs. Fondée sur une valorisation a posteriori des conditions qui caractérisent la période d’émergence du genre, qui s’étend de 1942 à 1957, l’authenticité insiste sur la continuité et la tradition. Ce discours, présent dès le milieu des années 1960, masque les aspects les plus modernes d’un genre qui, au moment où il émerge, n’est pas explicitement porteur de valeurs traditionnelles ou conservatrices. La voix country-western constitue un premier indice de modernité. La chanson country-western québécoise des années 1940 et 1950 structure dans un cadre musical des modificateurs paralinguistiques dont les deux plus caractéristiques, d’un point de vue générique, sont la nasalisation et le second mode de phonation. Véhicules de l’expressivité vocale, ces deux modificateurs du timbre sont coordonnés avec les paroles des chansons et avec la variation de paramètres musicaux, technologiques et phonétiques. Ils contribuent à la construction d’èthos spécifiques comme la tristesse, la solitude, la plainte et l’exubérance, et conservent dans le contexte discursif constitué par les enregistrements la signification expressive qu’on leur attribue dans la parole spontanée. C’est donc la voix parlée, quotidienne et ordinaire, qui fournit à l’auditeur le code culturel lui permettant d’en interpréter la signification. En ce sens, la chanson country-western incarne une certaine modernité populaire, redevable au code de la langue vulgaire partagée par le plus grand nombre. La modernité du country-western est aussi apparente dans sa popularité, qui se réalise à la fois dans son succès comme objet de consommation et dans sa proximité avec le public qui en détermine en partie le développement. Son recours particulier à la technologie, qui contribue à la création d’effets de spatialisation mais surtout à la mise en scène de l’intimité, le rattache aussi à la modernité. Enfin, le country-western témoigne d’une américanité certaine, assumée, et s’inscrit dans le déplacement du centre de gravité culturel, de l’Europe vers les États-Unis, qui marque la modernité. L’américanité du country-western, liée à ses origines mêmes, se renouvelle à la fin des années 1950 alors que le genre intègre le rock and roll. / This dissertation examines the relations between country-western music produced in Quebec between 1942 and 1957 and the concept of popular modernity. Drawing together musical and historical analysis, it explores the cultural significance of country-western at the beginning of the genre. The history of country-western music in Quebec began in 1942 when Roland Lebrun recorded his first songs. Paul Brunelle, Marcel Martel and Willie Lamothe soon followed with their own recordings in this style. These amateur singer-songwriters started out with the Compo Company and RCA Victor, the only two record companies who survived in Montreal during the Great Depression. With Rusticana, the first independent label to produce country-western music in 1958, the genre began its structuration and authenticity became a determinant value in country-western music. Continuity is a key concept to understand what Richard Peterson (1997) has called “fabricated authenticity”, which is indeed exemplified in the invented tradition that country-western created, from the 1960s onwards, on the basis of some of the conditions that characterized the first country-western. The discourse on authenticity, however, masks some of the more modern characteristics of the genre at the time of its birth. The country-western singing voice is one example, as artists use a variety of paralinguistic effects like nasalization and second mode of phonation (falsetto) in a way that can be seen as a stylized version of speech. Presenting the same expressive functions, these variations of timbre are coordinated with song lyrics, with musical and technological features, and with phonetic sounds to create symbolic representations of different emotions or èthos. These aesthetics based on everyday speech could be seen as a form of popular modernity: music for the people made by the people, but also for commercial success. Furthermore, country-western music used technology to create intimacy and spatialization effects. Its “américanité” was also marked and renewed when, in 1956 and 1957, country-western singers produced what can be considered some of the first rock'n'roll records in Quebec. These very modern features moderate the usual thesis about country-western’s traditionalism and conservatism.

Les personnes âgées forment une population qui connaît une forte croissance (Martel et Caron-Malenfant, 2007). Selon les données recueillies dans la présente étude, plusieurs de celles-ci sont intéressées par l’apprentissage du piano. Les aînés qui s’adonnent à l’apprentissage de la musique en retirent d’ailleurs des bénéfices sur les plans de la santé et du bien-être personnel (Creech, Hallam, McQueen et Varvarigou, 2013). De plus, les données colligées indiquent que la clientèle âgée des enseignants de piano est à la hausse et qu’il existe un manque de matériel pédagogique et de formation dans l’enseignement du piano aux aînés. La présente étude consiste donc en un projet de recherche de développement visant l’élaboration d’un guide pédagogique électronique destiné aux enseignants de piano travaillant avec des aînés ou souhaitant travailler avec cette clientèle. Rigoureusement conçu, ce guide a été élaboré à l’aide d’une adaptation des modèles de Van der Maren (1995; 2003), conçus pour l’élaboration d’un objet pédagogique. Répondant aux besoins recensés auprès de 118 enseignants de piano québécois et de huit aînés suivant des leçons de piano ou intéressés par cette activité d’apprentissage, ce guide pédagogique électronique est constitué de pièces de répertoire (partitions musicales et enregistrements sonores) adaptées aux préférences des personnes âgées québécoises et aux caractéristiques particulières de celles-ci, ainsi que de fiches et de capsules d’informations traitant des caractéristiques sensorielles, cognitives, psychomotrices, psychologiques et musicales des aînés. Il présente également des stratégies d’enseignement adaptées à l’enseignement du piano aux personnes âgées, des suggestions de lecture et de cahiers d’exercices techniques ainsi qu’une fiche d’évaluation de départ conçue afin d’aider les enseignants à identifier les besoins et le profil de leurs nouveaux élèves âgés. Le contenu et l’ergonomie web de l’outil conçu ont été validés par trois enseignants de piano expérimentés. / The aging population is constantly growing (Martel &amp; Caron-Malenfant, 2007). Data collected from the current study revealed a marked interest in piano learning for elderly students. Moreover, recent research indicates that « music continues to offer powerful potential for enhancing health and well-being in old age » (Creech, Hallam, McQueen &amp; Varvarigou, 2013). This study also revealed that piano teachers' older clientele is increasing while pedagogical materials and training resources for the elderly are not readily available. The objective of the current research project was to develop an online pedagogical guide specifically conceptualized for piano teachers who are working with, or who are willing to work with elderly students. This guide is based on an adaptation of the two research development models proposed by Van der Maren (1995; 2003) for the creation of pedagogical tools. The pedagogical guide was conceptualized to respond to the needs expressed by 188 piano teachers and by eight elderly piano students and prospective piano students. The guide consists of a piano repertoire (sheet music and recordings) that is adapted to older students' musical preferences and specific profiles, as well as accessible summaries of scientific articles and flash cards explaining sensorial, cognitive, psychomotor, psychological and musical characteristics of older piano students. The guide also includes teaching strategies adapted to elderly students, documents that explore a given subject in more detail, technical exercise book suggestions, and an evaluation form created to help teachers in identifying their elderly students’ needs and personal profiles. The content and the usability of the online pedagogical guide have been validated by three experienced piano teachers.

Cette thèse, complémentaire à deux récitals donnés dans le cadre de la réalisation d’un doctorat en études pédagogiques et pratiques musicales instrumentales, a comme fil conducteur l’enseignement collectif du trombone au Brésil et examine le trombone sous trois angles : historique, pédagogique et dans une perspective d’interprétation musicale. Bien que le sujet ne soit pas nouveau, le projet présente une réflexion, sur le trombone et son enseignement, encore inexplorée en raison principalement du répertoire sélectionné pour la présente étude. En effet, à notre connaissance, le répertoire populaire brésilien n’a jusqu’à maintenant pas fait l’objet d’arrangements musicaux écrits, en vue d’une exploitation non seulement en contexte de représentation publique, mais aussi avec des visées pédagogiques. Les morceaux choisis ont été retenus en raison de leur contenu, qui présente un vocabulaire musical lié à la culture régionale de quatre des cinq régions géographiques du Brésil. Dans le processus de réalisation du projet, les morceaux choisis ont fait l’objet d’arrangements dans la perspective qu’ils servent comme outil pédagogique pour la formation en trombone, en contexte de musique brésilienne. Les arrangements ont été conçus pour des ensembles réduits, où le trombone tient fréquemment le rôle de soliste. Ainsi, au-delà des concepts et techniques issus de la formation traditionnelle reçue, l’interprétation des arrangements a été une occasion pour le doctorant et soliste de présenter, à travers l'improvisation dans le langage de la musique populaire brésilienne, les résultats de processus développés au cours des deux dernières années. Les partitions sont annexées au texte de la thèse, afin d'illustrer certains propos et avec l’objectif de soutenir les recherches futures dans ce domaine. La thèse est divisée en quatre chapitres lesquels, bien que présentés séparément, possèdent entre eux d’étroites relations. Le premier chapitre donne un aperçu général de l'histoire de l'instrument, à travers les cinq siècles de son existence. Le deuxième chapitre s’articule autour de l'histoire de la musique brésilienne et de la présence du trombone au Brésil. Le troisième chapitre présente d’abord un aperçu de l’enseignement artistique et musical au Brésil au 20e et 21e siècle, poursuit en abordant la situation de l'enseignement du trombone au Brésil et aboutit à une proposition favorisant l’apprentissage du trombone en contexte d’enseignement musical collectif. Finalement, le quatrième chapitre fait un retour sur l’expérience musicale vécue dans le premier volet du projet, celui se rapportant à la pratique musicale instrumentale, établit des liens et trace des pistes dans une perspective d’enseignement et d’apprentissage du trombone. / This thesis, in conjunction with two recitals given toward fulfillment of a doctoral degree in pedagogical studies and instrumental musical practices, addresses collective teaching of the trombone in Brazil and examines the trombone from three perspectives: historical, educational and that of musical performance. Although the topic is not new, the project offers a reflection, on the trombone and how it is taught, unexplored mainly due to the repertoire selected for this study. Indeed, to our knowledge, written musical arrangements of the Brazilian popular repertoire have, until now, not been undertaken with a view both toward public performance as well as with pedagogical aims. The chosen pieces were selected because of their content, which presents a musical vocabulary related to the regional cultures of four of the five geographic regions of Brazil: the Northeast, the Midwest, the Southwest and the South. During the course of this project, the selected pieces were arranged such that they could serve as pedagogical tools for trombone training in the context of Brazilian music. The arrangements were designed for small ensembles, in which the trombone often plays the role of soloist. Thus, beyond the concepts and techniques gained through traditional training, performing the arrangements gave the doctoral student and soloist the opportunity to present, through the language of Brazilian popular music improvisation, the outcomes of processes developed over the past two years. Scores are appended to the text of the thesis, in order to illustrate certain remarks and with the goal of supporting future research in this area. The thesis is divided into four chapters, which, although presented separately, are closely related. The first chapter provides an overview of the history of the instrument through the five centuries of its existence. The second chapter focuses on the history of Brazilian music and on the presence of the trombone in Brazil. The third chapter begins with an overview of the artistic and musical education in Brazil in the twentieth and twenty-first centuries, then addresses the situation of teaching the trombone in Brazil, ultimately leading to a proposal to promote learning the trombone in the context of collective music education. Finally, the fourth chapter looks back at the musical experiences of the first part of this project, as an account of instrumental music practice, establishing links and following lines of thought with the perspective of teaching and learning the trombone.