Visual navigation is a ubiquitous yet complex task which is performed by many species for the purpose of survival. Although visual navigation is actively being studied within the robotics community, the determination of elemental constituents of a robust visual navigation system remains a challenge. Motion estimation is mistakenly considered as the sole ingredient to make a robust autonomous visual navigation system and therefore efforts are made to improve the accuracy of motion estimations. On the contrary, there are other factors which are as important as motion and whose absence could result in inability to perform seamless visual navigation such as the one exhibited by humans. Therefore, it is needed that a general model for a visual navigation system be devised which would describe it in terms of a set of elemental units. In this regard, a set of visual navigation elements (i.e. spatial memory, motion memory, scene geometry, context and scene semantics) are suggested as building blocks of a visual navigation system in this thesis. A set of methods are proposed which investigate the existence and role of visual navigation elements in a visual navigation system. A quantitative research methodology in the form of a series of systematic experiments is conducted on these methods. The thesis formulates, implements and analyzes the proposed methods in the context of visual navigation elements which are arranged into three major groupings; a) Spatial memory b) Motion Memory c) Manhattan, context and scene semantics. The investigations are carried out on multiple image datasets obtained by robot mounted cameras (2D/3D) moving in different environments. Spatial memory is investigated by evaluation of proposed place recognition methods. The recognized places and inter-place associations are then used to represent a visited set of places in the form of a topological map. Such a representation of places and their spatial associations models the concept of spatial memory. It resembles the humans’ ability of place representation and mapping for large environments (e.g. cities). Motion memory in a visual navigation system is analyzed by a thorough investigation of various motion estimation methods. This leads to proposals of direct motion estimation methods which compute accurate motion estimates by basing the estimation process on dominant surfaces. In everyday world, planar surfaces, especially the ground planes, are ubiquitous. Therefore, motion models are built upon this constraint. Manhattan structure provides geometrical cues which are helpful in solving navigation problems. There are some unique geometric primitives (e.g. planes) which make up an indoor environment. Therefore, a plane detection method is proposed as a result of investigations performed on scene structure. The method uses supervised learning to successfully classify the segmented clusters in 3D point-cloud datasets. In addition to geometry, the context of a scene also plays an important role in robustness of a visual navigation system. The context in which navigation is being performed imposes a set of constraints on objects and sections of the scene. The enforcement of such constraints enables the observer to robustly segment the scene and to classify various objects in the scene. A contextually aware scene segmentation method is proposed which classifies the image of a scene into a set of geometric classes. The geometric classes are sufficient for most of the navigation tasks. However, in order to facilitate the cognitive visual decision making process, the scene ought to be semantically segmented. The semantic of indoor scenes as well as semantic of the outdoor scenes are dealt with separately and separate methods are proposed for visual mapping of environments belonging to each type. An indoor scene consists of a corridor structure which is modeled as a cubic space in order to build a map of the environment. A “flash-n-extend” strategy is proposed which is responsible for controlling the map update frequency. The semantics of the outdoor scenes is also investigated and a scene classification method is proposed. The method employs a Markov Random Field (MRF) based classification framework which generates a set of semantic maps.

In Malawi, the new Biology curriculum draws from constructivist approaches. The Biology teaching syllabus emphasises the use of students’ experiences as a resource and the use of active student involvement strategies when teaching the curriculum to enhance conceptual understanding. However, inspection reports show that students are less actively involved in the lessons and the teachers rarely draw from their students’ prior experiences to enhance conceptual understanding. In this study, I investigated some factors affecting four teachers’ use of constructivist approaches when teaching the new biology curriculum in Malawi. Information was collected on the four teachers’ understandings of active student involvement in lessons and students’ experiences, the extent to which the four teachers involved students and built on students’ experiences to enable comprehension of science concepts in the Biology lessons and the factors that promoted/ hindered the use of these practices. Data was collected through lesson observations and interviews with teachers using an observation guide and an interview schedule respectively. Four Biology teachers, two from community secondary schools and two from conventional secondary schools were observed and interviewed. I found that all the four teachers understood students’ experiences as prior knowledge from previous school learning. The teachers understood active student involvement in relation to the type of schools they were teaching. For the teachers in community schools, involving students meant engaging them in group discussion while to the teachers in conventional schools it meant students doing experiments / practical work. Their teaching focused on giving information and little was done to develop students’ metacognitive abilities. According to the four teachers, pressure to cover the syllabus before the national examinations; lack of text books and laboratory equipment; lack of motivation among students; students’ backgrounds; and inadequate students’ fluency in the English language are some of the factors that affect their implementation of the constructivist approaches. Based on the lessons I observed, the teachers’ knowledge of subject matter and teachers’ understandings of the constructivism concept also affect the use of constructivist approaches when teaching Biology in Malawi.

Hume, no Tratado da Natureza Humana, afirma haver duas formas de o hábito atuar na produção de inferências, a saber, conforme princípios regulares e irregulares da imaginação. Em decorrência, estipula determinadas regras gerais para marcar a atuação do hábito no primeiro modo, restringindo a ela o espaço da causa e efeito. A intenção desta tese é investigar o estatuto dessas regras, bem como as suas consequências quanto ao estabelecimento das fronteiras entre a razão e a imaginação. Trata-se de questionar, inicialmente, qual é o parâmetro que permite uma separação, nos juízos, entre operações regulares e irregulares da imaginação, considerando-se que Hume mostra não haver uma justificativa racional para a relação de causa e efeito. Em contrapartida, pretende-se indicar em que medida uma nova noção de racionalidade experimental é configurada a partir da interposição desse novo critério, bem como discutir como é também a estabilização do agir do entendimento sobre a imaginação que se encontra no horizonte da normatividade instaurada pela regulação, via regras gerais do juízo. / In \"A Treatise of Human Nature\", Hume claims that there are two manners through which custom influences the production of inferences, namely, according to regular and irregular principles of imagination. Consequently, he stipulates certain general rules in order to point out the influence of custom on the first manner, circumscribing the realm of cause and effect to it. This thesis investigates these rules as well as their consequences regarding the establishment of the boundaries between reason and imagination. Considering that, according to Hume, there is not any racional justification to the cause-effect relationship, first we must question which is the parameter that allow us to separare, in reasoning, regular and irregular operation of the imagination. On the other hand, we intent to point in what extend a new notion of experimental rationality is constituted from the intervention of this new criteria. We also intent to discuss how the estabilization of understanding act works over imagination, which is placed in the range of normativity established by regulation, through the general rules of judgment.

Thesis advisor: Lillie R. Albert / To explore student mathematical self-efficacy and understanding of graphical data, this dissertation examines students solving real-world problems in their neighborhood, mediated by professional urban planning technologies. As states and schools are working on the alignment of the Common Core State Standards for Mathematics (CCSSM), traditional approaches to mathematics education that involves learning specific skills devoid of context will be challenged. For a student to be considered mathematically proficient according to the CCSSM, they must be able to understand mathematical models of real-world data, be proficient problem solvers and use appropriate technologies (tools) to be successful. This has proven to be difficult for all students--specifically for underrepresented students who have fallen behind in many of the Science, Technology, Engineering and Mathematics (STEM) fields. This mixed-method design involved survey and case-study research to collect and examine data over a two-year period. During the first year of this study, pre- and post-surveys using Likert-scale questions to all students in the urban planning project (n=62). During the two years, ten high school students' mathematical experiences while investigating urban planning projects in their own neighborhoods were explored through interviews, observations, and an examination of artifacts (eg. presentations and worksheets) in order to develop the case studies. Findings indicate that real-world mathematical tasks that are mediated by professional technologies influence both students' mathematical self-efficacy and understanding. Student self-efficacy was impacted by causing a shift in students beliefs about their own mathematical ability by having students interest increase through solving mathematical tasks that are rooted in meaningful, real-world contexts; students' belief that they can succeed in real-world mathematical tasks; and a shift in students' beliefs regarding the definition of `doing mathematics'. Results in light of mathematical understanding demonstrate that students' increased understanding was influenced by the ability to use multiple representations of data, making connections between the data and the physical site that was studied and the ability to communicate their findings to others. Implications for informal and formal learning, use of GIS in mathematics classrooms, and future research are discussed. / Thesis (PhD) — Boston College, 2013. / Submitted to: Boston College. Lynch School of Education. / Discipline: Teacher Education, Special Education, Curriculum and Instruction.

The research presented here shows the reflections of some International Baccalaureate (IB) Middle Years Programme (MYP) and Diploma students of a bilingual school on the attributes of the IB Learner Profile and the extent to which these can contribute to students being internationally minded. I approach the research after reviewing the concept of the self and using pragmatic methodology The research consisted of a questionnaire and semi-structured interview aimed at eliciting from the students their opinion about the extent to which they had equal command of the different attributes and, if they had differing abilities, if they were seeking to improve their command of the attribute(s) and who they though could help them do so. I was also interested in knowing if the students believed that the attributes of the IB Learner Profile helped them to be internationally minded, as suggested by the International Baccalaureate. The results suggest that most of the students believe that they have differing degrees of command of the attributes and that it is necessary for them to develop the ones that they feel less proficient in. They tend to believe that the attributes are developed partly at school, but also by their parents, themselves and others. They tend to believe that they are similar in terms of character to students that study at schools that do not offer the IB, but academically they are somewhat different. That said, they also tend to believe that the attributes of the IB Learner Profile help them to be internationally minded, particularly those of ‘open-minded’ and ‘communicator’, which coincide with the theoretical position of Castro et al (2013) and Singh and Qi (2013). While the school seems to play a significant part in developing the attributes, students seem to believe that they themselves, and parents, also have a key role to play.

Trata-se de investigar a atividade da compreensão em Hannah Arendt. Para tanto, assumiremos como horizonte desta pesquisa alguns aspectos do fenômeno totalitário. Nossa hipótese é que a compreensão, ao contrário de outras atividades mentais, está intimamente ligada ao mundo. Além disso, o compreender influencia a atitude dos indivíduos na medida em que os atrela aos acontecimentos. Por conseguinte, compreensão e acontecimento engendram uma nova experiência espaço-temporal, mediante a qual os fatos são desnaturalizados, abrindo-se, assim, à inspeção humana. A abertura operada pela compreensão não é, como poderíamos supor, de todo exterior ao homem. Segundo Arendt, o domínio do mundo e o domínio do pensamento comunicam suas experiências através das metáforas. Analisar o modo como o pensar se manifesta no mundo e como as experiências mundanas são apreendidas pelo pensamento mostrase fundamental para um bom entendimento da atividade da compreensão. / We intend to investigate the activity of understanding in Hannah Arendt. Therefore we will base this research on some aspects of the totalitarian phenomenon. Our hypothesis is that understanding, unlike other mental activities, is closely connected to the world. Furthermore, the activity of understanding influences the attitude of the men as approaches them to the events. Thus, understanding and events create a new space-time experience, by which facts are not naturalized, so the events are open to the human inspection. The opening created by the understanding is not, as we might suppose, external to men. According to Arendt, world and thought spaces communicate their experiences through metaphors. To analyze how the thinking manifests itself in the world and how the mundane experiences are apprehended by thought is crucial to reveal the implications of understanding.

Holistic scene understanding is one of the major goals in recent research of computer vision. Most popular recognition algorithms focus on semantic understanding and are incapable of providing the global depth information of the scene structure from the 2D projection of the world. Yet it is obvious that recovery of scene surface layout could be used to help many practical 3D-based applications, including 2D-to-3D movie re-production, robotic navigation, view synthesis, etc. Therefore, we identify scene geometric reasoning as the key problem of scene understanding. This PhD work makes a contribution to the reconstruction problem of 3D shape of scenes from monocular images. We propose an approach to recognise and reconstruct the geometric structure of the scene from a single image. We have investigated several typical scene geometries and built a few corresponding reference models in a hierarchical order for scene representation. The framework is set up based on the analysis of image statistical features and scene geometric features. Correlation is introduced to theoretically integrate these two types of features. Firstly, an image is categorized into one of the reference geometric models using the spatial pattern classi cation. Then, we estimate the depth pro le of the speci c scene by proposing an algorithm for adaptive automatic scene reconstruction. This algorithm employs speci cally developed reconstruction approaches for di erent geometric models. The theory and algorithms are instantiated in a system for the scene classi cation and visualization. The system is able to fi nd the best fi t model for most of the images from several benchmark datasets. Our experiments show that un-calibrated low-quality monocular images could be e fficiently and realistically reconstructed in simulated 3D space. By our approach, computers could interpret a single still image as its underlying geometry straightforwardly, avoiding usual object occlusion, semantic overlapping and defi ciency problems.

Through a singular case study, this research enquiry seeks to explore teacher perceptions about the development of a concept-based curriculum program (called as the Conceptual Curriculum by the school), in the context of an International Baccalaureate (IB) World school in the Middle East, and the intended and unintended outcomes of the initiative. The study employs Bernstein’s (1975) theories of classification and framing, and curriculum recontextualization, as an analytical framework to interpret findings. The study is informed through methods such as reading and analyzing of curriculum documents, conducting semi-structured interviews, and the distribution of a web-based questionnaire to teachers. Findings in this research inquiry revealed that, though teachers expressed the experience of creating and delivering the Conceptual Curriculum as sometimes being challenging and frustrating, a vast majority of the teachers prefer a flexible curriculum framework versus a prescriptive curriculum. However, findings also revealed that, though teachers seem to enjoy the freedom and flexibility of working with broad curricular frameworks as opposed to prescriptive curricula, there seem to be some fundamental questions pertinent to curriculum recontextualization remaining unanswered, for which perhaps teachers seek answers from qualified curriculum development personnel. Findings reveal that when broad curricular frameworks get recontextualized, the lack of consensus amongst teachers on what counts as essential knowledge is often a matter of concern. Findings reveal that in curriculum recontextualization, when having to negotiate between a “multiplicity of pedagogic fields” (Cambridge, 2011, p. 129) teachers seem to be inherently aligning to something that is a “crystal clear benchmark” such as the International Baccalaureate Diploma Programme (IB DP), as opposed to something that is more flexible and open-ended such as the International Baccalaureate Primary Years Programme (IB PYP). The disciplinary focus of the Conceptual Curriculum and the tendency of teachers to align more towards the IB DP rather than the IB PYP (even in lower grades such as 7 and 8) has thus resulted in a quick transition from the “weakly classified” (Bernstein 1971, p.49) inter-disciplinary IB PYP curriculum to a “strongly classified” (Bernstein 1971, p.49) Conceptual Curriculum with disciplinary focus. Findings from this study reveal that teachers see the value and purpose in teaching for conceptual understanding, but this, when coupled with having to choose curriculum content and developing a coherent curriculum has made the experience both challenging and burdensome for them. Findings also reveal that practical agendas of the school, such as addressing limited time and staffing issues assume priority over lofty ideals when the curriculum is recontextualized, thereby indicating that school-based curriculum initiatives lose rigor and form, in the cracks of everyday practice. Findings in this study thus suggest that when teachers are offered the possibility of working with flexible curricular frameworks, realities of everyday practice take over. This often leads to teachers self-prescribing the curriculum, thereby making the process self-mandated, which in effect defeats the very purpose of the school-based curriculum development initiative undertaken.

With the growth of online media, surveillance and mobile cameras, the amount and size of video databases are increasing at an incredible pace. For example, YouTube reported that over 400 hours of video are uploaded every minute to their servers. Arguably, people are the most important and interesting subjects of such videos. The computer vision community has embraced this observation to validate the crucial role that human action recognition plays in building smarter surveillance systems, semantically aware video indexes and more natural human-computer interfaces. However, despite the explosion of video data, the ability to automatically recognize and understand human activities is still somewhat limited. In this work, I address four different challenges at scaling up action understanding. First, I tackle existing dataset limitations by using a flexible framework that allows continuous acquisition, crowdsourced annotation, and segmentation of online videos, thus, culminating in a large-scale, rich, and easy-to-use activity dataset, known as ActivityNet. Second, I develop an action proposal model that takes a video and directly generates temporal segments that are likely to contain human actions. The model has two appealing properties: (a) it retrieves temporal locations of activities with high recall, and (b) it produces these proposals quickly. Thirdly, I introduce a model, which exploits action-object and action-scene relationships to improve the localization quality of a fast generic action proposal method and to prune out irrelevant activities in a cascade fashion quickly. These two features lead to an efficient and accurate cascade pipeline for temporal activity localization. Lastly, I introduce a novel active learning framework for temporal localization that aims to mitigate the data dependency issue of contemporary action detectors. By creating a large-scale video benchmark, designing efficient action scanning methods, enriching approaches with high-level semantics for activity localization, and an effective strategy to build action detectors with limited data, this thesis is making a step closer towards general video understanding.

This thesis investigates to what extent do cognitive biases a ect human understanding of interpretable machine learning models, in particular of rules discovered from data. Twenty cognitive biases (illusions, e ects) are analysed in detail, including identi cation of possibly e ective debiasing techniques that can be adopted by designers of machine learning algorithms and software. This qualitative research is complemented by multiple experiments aimed to verify, whether, and to what extent, do selected cognitive biases in uence human understanding of actual rule learning results. Two experiments were performed, one focused on eliciting plausibility judgments for pairs of inductively learned rules, second experiment involved replication of the Linda experiment with crowdsourcing and two of its modi cations. Altogether nearly 3.000 human judgments were collected. We obtained empirical evidence for the insensitivity to sample size e ect. There is also limited evidence for the disjunction fallacy, misunderstanding of and , weak evidence e ect and availability heuristic. While there seems no universal approach for eliminating all the identi ed cognitive biases, it follows from our analysis that the e ect of many biases can be ameliorated by making rule-based models more concise. To this end, in the second part of thesis we propose a novel machine learning framework which postprocesses rules on the output of the seminal association rule classi cation algorithm CBA [Liu et al, 1998]. The framework uses original undiscretized numerical attributes to optimize the discovered association rules, re ning the boundaries of literals in the antecedent of the rules produced by CBA. Some rules as well as literals from the rules can consequently be removed, which makes the resulting classi er smaller. Benchmark of our approach on 22 UCI datasets shows average 53% decrease in the total size of the model as measured by the total number of conditions in all rules. Model accuracy remains on the same level as for CBA.