Thesis (Ed.M.)--Boston University / PLEASE NOTE: Boston University Libraries did not receive an Authorization To Manage form for this thesis or dissertation. It is therefore not openly accessible, though it may be available by request. If you are the author or principal advisor of this work and would like to request open access for it, please contact us at open-help@bu.edu. Thank you. / 2031-01-01

A self-­‐learning Audio Player was built to learn users habits by analyzing operations the user does when listening to music. The self-­‐learning component is intended to provide a better music experience for the user by generating a special playlist based on the prediction of users favorite songs. The rough set core characteristics are used throughout the learning process to capture the dynamics of changing user interactions with the audio player. The engine is evaluated by simulation data. The simulation process ensures the data contain specific predetermined patterns. Evaluation results show the predictive power and stability of the hybrid engine for learning a users habits and the increased intelligence achieved by combining rough sets and NN when compared with using NN by itself.

Yes / This paper presents the design of a multi-agent framework that aids engineers in the adaptation of modular production systems. The framework includes general implementations of agents and other software components for self-learning and adaptation, sensor data analysis, system modelling and simulation, as well as human-computer interaction. During an adaptation process, operators make changes to the production system, in order to increase capacity or manufacture a product variant. These changes are automatically captured and evaluated by the framework, building an experience base of adjustments that is then used to infer adaptation knowledge. The architecture of the framework consists of agents divided in two layers: the agents in the lower layer are associated with individual production modules, whereas the agents in the higher layer are associated with the entire production line. Modelling, learning, and adaptations can be performed at both levels, using a semantic model to specify the structure and capabilities of the production system. An evaluation of a prototype implementation has been conducted on an industrial assembly system. The results indicate that the use of the framework in a typical adaptation process provides a significant reduction in time and resources required. / This work was supported in part by the European Commission [grant agreement n. 314762].

In this thesis I have examined the benefits of working with a help of self-recording. I wanted to experiment self-monitoring with different working methods to improve my practicing skills as well my performing skills as I prepared for my examination concert. This process consisted of a lot of recording, listening and practicing and repeating this cycle numerous times. I wanted to implement different practice methods and reflect on different aspects of playing the viola. This thesis is written from a violist point of view. The main focus of this work is audio recording as a tool in self-monitoring practice. I have been using two main methods in reviewing the audio material gathered from practice sessions:  1) time between recording and reviewing the material and 2) recording, analyzing and practicing in line with the recordings within a practice session. I wanted to take self-recording process into more regular use because I see it as an essential part of the self-teaching process. A music student has to go through a lot of practicing hours and most of these are spent alone in a practice room. Some of this time is wasted and misused in learning unwanted habits. I wanted to learn to practice in the most deliberate way and use my practice hours as effectively as possible by structuring my practice in self-teaching phases and putting the emphasis on self-monitoring. / <p>CONCERT REPERTOAR</p><p><strong>C. Stamitz</strong>: Viola Concerto in D Major, Op. 1*</p><p>Allegro</p><p><strong>I. Stravinsky</strong>: Elegie</p><p><strong>R. Schumann</strong>: Märchenbilder, Op. 113**</p><p><em>I. Nicht Schnell</em></p><p><em>II. Lebhaft</em></p><p><em>III. Rasch</em></p><p><em>IV. Langsam, mit melancholischen Ausdrück</em></p><p><strong>B. Bartok</strong>: Concerto for Viola and Orchester, Sz. 120, BB 128*</p><p>I. Moderato</p><p>Pianist:</p><p>*= Erik Lanninger</p><p>**= Eeva Tapanen</p><p></p>

Recently in pedagogical literature more and more attention to adult education and whole life studies is given. After beginning of the Movement, searches was started for new ways how to wake up a teacher for refusing authoritarianism, dictatorship working methods, changing attitude to work, knowing children psychology, evolution singularities, improving relations with colleagues and pupils, to feel self value, dignity, national self-respect, being self-sufficient educator. Ph. D. M.Lukšienė about 1991 education reform was writing: “the most important member of changeover – teacher […]. We will not get to feet without growing and educating person. […]. Education and all school system with all nurture meaning institutions appear with very important role”. Educator perfection is not distinguishable from school perfection, it means youth generation future depends on that too. Big part of the professional perfection takes pedagogical and personal self-help, which is not all the time stimulated by internal motivation and positive emotions. Usually educator is improving himself only as much as it is a needed for qualification enhance. There are cases, when certificates are bought and methodical meetings are only formality. Plus, not every time organized region events satisfy educators’ needs. I am primary school teacher, so I will examine my colleagues – primary school teachers’ pedagogical self-help singularity and specificity. In my work there will be pedagogical self-help... [to full text]

Recent advances in large neural network-style models have demonstrated great performance in various applications, such as image generation, question answering, and audio classification. However, these deep and high-capacity models require a large amount of labeled data to function properly, rendering them inapplicable in many real-world scenarios. This dissertation focuses on the development and evaluation of advanced machine learning algorithms to solve the following research questions: (1) How to learn novel classes with limited labeled data, (2) How to adapt a large pre-trained model to the target domain if only unlabeled data is available, (3) How to boost the performance of the few-shot learning model with unlabeled data, and (4) How to utilize limited labeled data to learn new classes without the training data in the same domain. First, we study few-shot learning in text classification tasks. Meta-learning is becoming a popular approach for addressing few-shot text classification and has achieved state-of-the-art performance. However, the performance of existing approaches heavily depends on the interclass variance of the support set. To address this problem, we propose a TART network for few-shot text classification. The model enhances the generalization by transforming the class prototypes to per-class fixed reference points in task-adaptive metric spaces. In addition, we design a novel discriminative reference regularization to maximize divergence between transformed prototypes in task-adaptive metric spaces to improve performance further. In the second problem we focus on self-learning in cross-lingual transfer task. Our goal here is to develop a framework that can make the pretrained cross-lingual model continue learning the knowledge with large amount of unlabeled data. Existing self-learning methods in crosslingual transfer tasks suffer from the large number of incorrectly pseudo-labeled samples used in the training phase. We first design an uncertainty-aware cross-lingual transfer framework with pseudo-partial-labels. We also propose a novel pseudo-partial-label estimation method that considers prediction confidences and the limitation to the number of candidate classes. Next, to boost the performance of the few-shot learning model with unlabeled data, we propose a semi-supervised approach for few-shot semantic segmentation task. Existing solutions for few-shot semantic segmentation cannot easily be applied to utilize image-level weak annotations. We propose a class-prototype augmentation method to enrich the prototype representation by utilizing a few image-level annotations, achieving superior performance in one-/multi-way and weak annotation settings. We also design a robust strategy with softmasked average pooling to handle the noise in image-level annotations, which considers the prediction uncertainty and employs the task-specific threshold to mask the distraction. Finally, we study the cross-domain few-shot learning in the semantic segmentation task. Most existing few-shot segmentation methods consider a setting where base classes are drawn from the same domain as the new classes. Nevertheless, gathering enough training data for meta-learning is either unattainable or impractical in many applications. We extend few-shot semantic segmentation to a new task, called Cross-Domain Few-Shot Semantic Segmentation (CD-FSS), which aims to generalize the meta-knowledge from domains with sufficient training labels to low-resource domains. Then, we establish a new benchmark for the CD-FSS task and evaluate both representative few-shot segmentation methods and transfer learning based methods on the proposed benchmark. We then propose a novel Pyramid-AnchorTransformation based few-shot segmentation network (PATNet), in which domain-specific features are transformed into domain-agnostic ones for downstream segmentation modules to fast adapt to unseen domains. / Doctor of Philosophy / Nowadays, deep learning techniques play a crucial role in our everyday existence. In addition, they are crucial to the success of many e-commerce and local businesses for enhancing data analytics and decision-making. Notable applications include intelligent transportation, intelligent healthcare, the generation of natural language, and intrusion detection, among others. To achieve reasonable performance on a new task, these deep and high-capacity models require thousands of labeled examples, which increases the data collection effort and computation costs associated with training a model. Moreover, in many disciplines, it might be difficult or even impossible to obtain data due to concerns such as privacy and safety. This dissertation focuses on learning with limited labeled data in natural language processing and computer vision tasks. To recognize novel classes with a few examples in text classification tasks, we develop a deep learning-based model that can capture both cross- task transferable knowledge and task-specific features. We also build an uncertainty-aware self-learning framework and a semi-supervised few-shot learning method, which allow us to boost the pre-trained model with easily accessible unlabeled data. In addition, we propose a cross-domain few-shot semantic segmentation method to generalize the model to different domains with a few examples. By handling these unique challenges in learning with limited labeled data and developing suitable approaches, we hope to improve the efficiency and generalization of deep learning methods in the real world.

Robots are destined to move beyond the caged factory floors towards domains where they will be interacting closely with humans. They will encounter highly varied environments, scenarios and user demands. As a result, programming robots after deployment will be an important requirement. To address this challenge, the field of Learning from Demonstration (LfD) emerged with the vision of programming robots through demonstrations of the desired behavior instead of explicit programming. The field of LfD within robotics has been around for more than 30 years and is still an actively researched field. However, very little research is done on the implications of having a non-robotics expert as a teacher. This thesis aims to bridge this gap by developing learning from demonstration algorithms and interaction paradigms that allow non-expert people to teach robots new skills. The first step of the thesis was to evaluate how non-expert teachers provide demonstrations to robots. Keyframe demonstrations are introduced to the field of LfD to help people teach skills to robots and compared with the traditional trajectory demonstrations. The utility of keyframes are validated by a series of experiments with more than 80 participants. Based on the experiments, a hybrid of trajectory and keyframe demonstrations are proposed to take advantage of both and a method was developed to learn from trajectories, keyframes and hybrid demonstrations in a unified way. A key insight from these user experiments was that teachers are goal oriented. They concentrated on achieving the goal of the demonstrated skills rather than providing good quality demonstrations. Based on this observation, this thesis introduces a method that can learn actions and goals from the same set of demonstrations. The action models are used to execute the skill and goal models to monitor this execution. A user study with eight participants and two skills showed that successful goal models can be learned from non- expert teacher data even if the resulting action models are not as successful. Following these results, this thesis further develops a self-improvement algorithm that uses the goal monitoring output to improve the action models, without further user input. This approach is validated with an expert user and two skills. Finally, this thesis builds an interactive LfD system that incorporates both goal learning and self-improvement and evaluates it with 12 naive users and three skills. The results suggests that teacher feedback during experiments increases skill execution and monitoring success. Moreover, non-expert data can be used as a seed to self-improvement to fix unsuccessful action models.

In the automotive industry test drives are conducted during the development of new vehicle models or as a part of quality assurance for series vehicles. During the test drives, data is recorded for the use of fault analysis resulting in millions of data points. Since multiple vehicles are tested in parallel, the amount of data that is to be analysed is tremendous. Hence, manually analysing each recording is not feasible. Furthermore the complexity of vehicles is ever-increasing leading to an increase of the data volume and complexity of the recordings. Only by effective means of analysing the recordings, one can make sure that the effort put in the conducting of test drives pays off. Consequently, effective means of test drive analysis can become a competitive advantage. This Thesis researches ways to detect unknown or unmodelled faults in recordings from test drives with the following two aims: (1) in a data base of recordings, the expert shall be pointed to potential errors by reporting anomalies, and (2) the time required for the manual analysis of one recording shall be shortened. The idea to achieve the first aim is to learn the normal behaviour from a training set of recordings and then to autonomously detect anomalies. The one-class classifier “support vector data description” (SVDD) is identified to be most suitable, though it suffers from the need to specify parameters beforehand. One main contribution of this Thesis is a new autonomous parameter tuning approach, making SVDD applicable to the problem at hand. Another vital contribution is a novel approach enhancing SVDD to work with multivariate time series. The outcome is the classifier “SVDDsubseq” that is directly applicable to test drive data, without the need for expert knowledge to configure or tune the classifier. The second aim is achieved by adapting visual data mining techniques to make the manual analysis of test drives more efficient. The methods of “parallel coordinates” and “scatter plot matrices” are enhanced by sophisticated filter and query operations, combined with a query tool that allows to graphically formulate search patterns. As a combination of the autonomous classifier “SVDDsubseq” and user-driven visual data mining techniques, a novel, data-driven, semi-autonomous approach to detect unmodelled faults in recordings from test drives is proposed and successfully validated on recordings from test drives. The methodologies in this Thesis can be used as a guideline when setting up an anomaly detection system for own vehicle data.

Submitted by Filipe dos Santos (fsantos@pucsp.br) on 2017-10-23T12:22:00Z No. of bitstreams: 1 Letícia Ferreira Frigo.pdf: 1475126 bytes, checksum: a0366034bcf56a5b72fe550bfe772a50 (MD5) / Made available in DSpace on 2017-10-23T12:22:00Z (GMT). No. of bitstreams: 1 Letícia Ferreira Frigo.pdf: 1475126 bytes, checksum: a0366034bcf56a5b72fe550bfe772a50 (MD5) Previous issue date: 2017-09-06 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / The present research deals with the use of digital technologies in the classroom as an auxiliary resource in the teaching-learning process, aiming at the individual and / or collective cognitive extension of subjects guided by interactivity and the adoption of collaborative learning practices: hybrid teaching and connectivism. The corpus of this research is anchored in the use of these technologies in the construction of the student's autonomy, making him the co-author of his learning, from the information search process to the content transformation, through filtering and data selection. Within this link, the following questions are raised: [1] Are the procedures, which presuppose interactive bilaterality, adopted by the practice of hybrid teaching, and establishe the use of such technologies in the classroom, to be applicable? [2] Do connectivist practices that seek to expand content circulating on the web through data sharing really form networks of interests in which the subjects involved exchange discoveries? [3] What are the reasons why many schools and teachers reject technological immersion in school, discouraging and / or prohibiting the use of such equipment in schools? The hypotheses drawn during this research point to the need to define how these processes become possible; and to this end, the topics addressed to digital learning highlight some of these practices, such as the search for information, through rhizomatic navigation and hypertextual and / or hypermedia reading, and the development of self-learning practices. The research highlights the role of the teacher in face of this new panorama and his current attributions regarding the methodologies for the insertion of the subjects in the cyberspace. The objective of this work is justified in the use of digital equipment as auxiliary tools to promote the cognitive expansion of subjects and to increase collective intelligence through playful and transformative practices. The result of the research, noting these two intellectual effects through the use of these technologies and the large-scale exclusion of society, which is on the margins of this process, advocates the need to include more subjects in these innovations and democratize the use of digital technologies as way to reduce social differences generated by educational discrepancies in our society. Research indicates that matching the learning opportunities will reduce the social gulf among the students; Therefore, the adoption of collaborative practices with the use of digital technologies promotes inclusion in the learning spheres, potentially improving their socioeconomic conditions / A presente pesquisa versa sobre o uso das tecnologias digitais em sala de aula como recurso auxiliar no processo de ensino-aprendizagem, visando à ampliação cognitiva individual e/ou coletiva dos sujeitos pautada pela interatividade e pela adoção de práticas colaborativas de aprendizagem: ensino híbrido e conectivismo. O corpus desta pesquisa ancora-se no uso dessas tecnologias na construção da autonomia do aluno, tornando-o coautor de sua aprendizagem, desde o processo de busca de informações até a transformação dos conteúdos, passando pela filtragem e seleção dos dados. Dentro desse enlace, são levantadas as seguintes questões: [1] Os procedimentos, que pressupõem bilateralidade interativa, adotados pela prática do ensino híbrido, e estabelecem o uso de tais tecnologias em sala de aula seriam de fato aplicáveis? [2] As práticas conectivistas que visam ampliação dos conteúdos circulantes na web a partir do compartilhamento de dados, realmente formam redes de interesses nas quais os sujeitos envolvidos trocam descobertas? [3] Quais os motivos que levam muitas escolas e professores a rechaçarem a imersão tecnológica no âmbito escolar, desestimulando e ou proibindo o uso destes equipamentos nas escolas? As hipóteses traçadas durante essa pesquisa apontam para a necessidade de definir como esses processos tornam-se possíveis; e, para tanto, os tópicos destinados à aprendizagem digital ressaltam algumas dessas práticas, tais como a busca pela informação, por meio da navegação rizomática e de leituras hipertextuais e/ou hipermidiáticas, e o desenvolvimento de práticas de autoaprendizagem. A pesquisa ressalta o papel do professor diante desse novo panorama e de suas atuais atribuições quanto às metodologias para a inserção dos sujeitos no ciberespaço. O objetivo deste trabalho justifica-se no uso de equipamentos digitais como ferramentas auxiliares à promoção da ampliação cognitiva dos sujeitos e ao aumento da inteligência coletiva por meio de práticas lúdicas e transformadoras. O resultado da pesquisa, constatando esses dois efeitos intelectivos por meio do uso dessas tecnologias e a exclusão de grande camada da sociedade, que se encontra à margem desse processo, advoga a necessidade de incluir mais sujeitos nessas inovações e democratizar o uso das tecnologias digitais como forma de reduzir diferenças sociais geradas pelas discrepâncias educacionais em nossa sociedade. A pesquisa indica que se igualando as oportunidades de aprendizagem, diminuiri-se-a o abismo social entre os sujeitos; logo, a adoção de práticas colaborativas, com o uso das tecnologias digitais, promove inclusão nas esferas de aprendizagem, melhorando, potencialmente, suas condições socioeconômicas

Atrial fibrillation (AF) is a common cardiac arrhythmia that requires extensive medical and pharmaceutical management. The coagulation antagonist warfarin is commonly prescribed to reduce AF-associated stroke. Although warfarin effectively mediates thromboembolitic risk, its management is complex as many factors influence its therapeutic range including: genetics, diet, medication, and herbal and dietary supplement (HDS) interactions. Lack of patient knowledge regarding these factors contributes to poor patient outcomes. With the emerging epidemic of AF, readily available educational tools are necessary to improve patient outcomes while reducing clinician burden. The purpose of this study was to develop both a self-learning, DVD-based and one-on-one education program to educate patients with atrial fibrillation about the risks of HDS-warfarin interactions and to compare education method efficacy in AF disease management. This study found patients lack knowledge regarding HDS-warfarin management, and both DVD-based and one-on-one education models could increase patient knowledge regarding HDS-warfarin factors. It is hypothesized this education method may be employed to further educate chronic disease populations about essential disease-associated factors to improve outcomes while reducing clinical burdens.