No description available.

This thesis is about the dynamic generalisation of continuous action spaces in reinforcement learning problems. The standard Reinforcement Learning (RL) account provides a principled and comprehensive means of optimising a scalar reward signal in a Markov Decision Process. However, the theory itself does not directly address the imperative issue of generalisation which naturally arises as a consequence of large or continuous state and action spaces. A current thrust of research is aimed at fusing the generalisation capabilities of supervised (and unsupervised) learning techniques with the RL theory. An example par excellence is Tesauro’s TD-Gammon. Although much effort has gone into researching ways to represent and generalise over the input space, much less attention has been paid to the action space. This thesis first considers the motivation for learning real-valued actions, and then proposes a set of key properties desirable in any candidate algorithm addressing generalisation of both input and action spaces. These properties include: Provision of adaptive and online generalisation, adherence to the standard theory with a central focus on estimating expected reward, provision for real-valued states and actions, and full support for a real-valued discounted reward signal. Of particular interest are issues pertaining to robustness in non-stationary environments, scalability, and efficiency for real-time learning in applications such as robotics. Since exploring the action space is discovered to be a potentially costly process, the system should also be flexible enough to enable maximum reuse of learned actions. A new approach is proposed which succeeds for the first time in addressing all of the key issues identified. The algorithm, which is based on the ubiquitous self-organising map, is analysed and compared with other techniques including those based on the backpropagation algorithm. The investigation uncovers some important implications of the differences between these two particular approaches with respect to RL. In particular, the distributed representation of the multi-layer perceptron is judged to be something of a double-edged sword offering more sophisticated and more scalable generalising power, but potentially causing problems in dynamic or non-equiprobable environments, and tasks involving a highly varying input-output mapping. The thesis concludes that the self-organising map can be used in conjunction with current RL theory to provide real-time dynamic representation and generalisation of continuous action spaces. The proposed model is shown to be reliable in non-stationary, unpredictable and noisy environments and judged to be unique in addressing and satisfying a number of desirable properties identified as important to a large class of RL problems.

Traditional supervised visual recognition methods require a great number of annotated examples for each concerned class. The collection and annotation of visual data (e.g., images and videos) could be laborious, tedious and time-consuming when the number of classes involved is very large. In addition, there are such situations where the test instances are from novel classes for which training examples are unavailable in the training stage. These issues can be addressed by zero-shot learning (ZSL), an emerging machine learning technique enabling the recognition of novel classes. The key issue in zero-shot visual recognition is the semantic gap between visual and semantic representations. We address this issue in this thesis from three different perspectives: visual representations, semantic representations and the learning models. We first propose a novel bidirectional latent embedding framework for zero-shot visual recognition. By learning a latent space from visual representations and labelling information of the training examples, instances of different classes can be mapped into the latent space with the preserving of both visual and semantic relatedness, hence the semantic gap can be bridged. We conduct experiments on both object and human action recognition benchmarks to validate the effectiveness of the proposed ZSL framework. Then we extend the ZSL to the multi-label scenarios for multi-label zero-shot human action recognition based on weakly annotated video data. We employ a long short term memory (LSTM) neural network to explore the multiple actions underlying the video data. A joint latent space is learned by two component models (i.e. the visual model and the semantic model) to bridge the semantic gap. The two component embedding models are trained alternately to optimize the ranking based objectives. Extensive experiments are carried out on two multi-label human action datasets to evaluate the proposed framework. Finally, we propose alternative semantic representations for human actions towards narrowing the semantic gap from the perspective of semantic representation. A simple yet effective solution based on the exploration of web data has been investigated to enhance the semantic representations for human actions. The novel semantic representations are proved to benefit the zero-shot human action recognition significantly compared to the traditional attributes and word vectors. In summary, we propose novel frameworks for zero-shot visual recognition towards narrowing and bridging the semantic gap, and achieve state-of-the-art performance in different settings on multiple benchmarks.

No description available.

The thesis seeks to develop an account of collaborative activities within the framework of ecological realism—an approach to psychology developed by James J. Gibson in the course of work on visual perception. Two main questions are addressed; one ontological, and one methodological. The ontological question is: given that collaborative activities take place within an environment, what kinds of structure must this environment contain? The response emphasizes the importance of relations which exist between entities, and which connect a given perceiver-actor with the other objects and individuals in its surroundings, and with the relations between those entities. It is held that activities take place within a field of relations. This description draws on the radical empiricist doctrine that relations are real, are external, and are directly perceivable. The present proposal insists that, in addition to being directly perceivable, relations can also be directly acted upon: throwing a ball for a dog is acting on a relation between dog and ball in space. The relational field account of collaboration naturally extends to an account of speaking: people, through their history of acting in an environment populated by other speakers, come to stand in a set of relations with objects and events around them, and these relations can be directly acted upon by others through the use of verbal actions. Verbal actions serve to direct the attention of others to relevant aspects of the environment, and this allows us as speakers to coordinate and manage one another’s activity. The methodological question is this: granting that the environment may be structured as a field of relations, how are we to conduct our empirical investigations, such that we can ask precise questions which lead to useful insights about how a given collaborative activity is carried out in practice? The central issue here concerns the concept of the task. Psychologists are in the habit of using this term quite loosely, to denote the actions of an individual or a group, in a laboratory or outside. This creates confusion in discussions of collaborative phenomena: who is the agent of a ‘collaborative task’? The definition offered here states that a task is a researcher-defined unit of study that corresponds to a change in the structure of the environment that has a characteristic pattern and that is meaningful from the first-person perspective of a particular actor. On this definition, the task is a tool that allows ecological psychologists to carve up the problem space into specific, tractable questions; the task is the equivalent of the cognitivist’s mental module. Task-oriented psychology encourages us to ask the question: which specific resources is the individual making use of in controlling this particular activity? The methodology is developed through an examination of the alarm calling behaviour of vervet monkeys, which is explained in terms of actions on the relational field, and through an analysis of corpus data from a laboratory-based collaborative assembly game. The relational field model promises to provide a way of studying social and collaborative activities on ecological realist principles. The concluding chapter identifies two particular areas in which the model might fruitfully be developed: in the study of learning, and in the theory of designing objects and spaces for interaction.

Background: Mind-body dualisms create particular difficulties for researching and justifying learning and knowledge within PE practices. These issues are compounded in the UK by prevailing cognitivistic ideas of education, knowledge and learning. Crum (1993) suggests reconceptualising PE as movement culture as a potential solution to the limitations created by dualistic positions within education. How knowledge and learning within movement culture is positioned, however, was left underdeveloped by Crum. The aim of this thesis is to explore an embodied, action position on knowledge and learning, as a potential solution to this issue. Purpose: This thesis is driven by two purposes. The first; to examine and discuss how John Dewey’s theorising of knowledge and learning within experience provides a theoretical position on knowledge and learning within movement culture. The second; to utilise this position to explore how pupils’ and teachers’ actions within primary PE lessons constitute and negotiate the movement cultures within their school. Findings: In adopting a position which dissolves mind-body dualisms, movement culture allows the practical work of PE lessons to be considered as contexts of knowledge production. This opens up our understanding of different ways of knowing in PE through pupils’ epistemological ‘action-in-PE-settings’. Rather than creating another hybrid of educational ideology by objectifying what to ‘do’ or ‘know’, movement culture keeps the ‘who’ of participation in PE practice in view. Such a position is achieved because pupils are seen as ‘coming to know’ through their immediate and continuous experiences of sports and physical activities both in PE and beyond the school gates. By dissolving traditional dualisms within educational ideology, movement culture allows ideologies and assumptions about learning in PE to be decoded and managed. It also provides a framework to explore subject-matter for learning and analyses some of the disconnections which exist within PE practice. Conclusions: Reconceptualising PE as movement culture is not intended to create a logic of practice to which I claim PE should ascribe. In this thesis, movement culture offers a position from which to consider the continuity between PE and pupils’ lives within and outside of the school gates. Such a standpoint can challenge our ideas as to what subject-matter could be within PE and the possibilities of learning outcomes other than those that focus on performance sport or bodily training for fitness. From a research perspective questions arise in relation to understanding very young pupils’ experiences of knowing within PE and how learning and knowledge are embodied across other subject areas. Addressing such questions may help to support new understandings of learning and knowledge within schools that are concurrent with developing new methodologies and research tools. These may in turn support the continuing development of pedagogical practices.

Humans are unique in their ability to flexibly and rapidly adapt their behaviour and select courses of action that lead to future reward. Several ‘component processes’ must be implemented by the human brain in order to facilitate this behaviour. This thesis examines two such components; (i) the neural substrates supporting action selection during value- guided choice using magnetoencephalography (MEG), and (ii) learning the value of environmental stimuli and other people’s actions using functional magnetic resonance imaging (fMRI). In both situations, it is helpful to formally model the underlying component process, as this generates predictions of trial-to-trial variability in the signal from a brain region involved in its implementation. In the case of value-guided action selection, a biophysically realistic implementation of a drift diffusion model is used. Using this model, it is predicted that there are specific times and frequency bands at which correlates of value are seen. Firstly, there are correlates of the overall value of the two presented options, and secondly the difference in value between the options. Both correlates should be observed in the local field potential, which is closely related to the signal measured using MEG. Importantly, the content of these predictions is quite distinct from the function of the model circuit, which is to transform inputs relating to the value of each option into a categorical decision. In the case of social learning, the same reinforcement learning model is used to track both the value of two stimuli that the subject can choose between, and the advice of a confederate who is playing alongside them. As the confederate advice is actually delivered by a computer, it is possible to keep prediction error and learning rate terms for stimuli and advice orthogonal to one another, and so look for neural correlates of both social and non-social learning in the same fMRI data. Correlates of intentional inference are found in a network of brain regions previously implicated in social cognition, notably the dorsomedial prefrontal cortex, the right temporoparietal junction, and the anterior cingulate gyrus.

The dissertation substantiates the efficacy of evidence-based teacher’s activity, teaching students with emotional and behavioural disorders in the general education school. The efficacy of evidence-based teacher’s activity is analysed according to two criteria: teacher’s learning-in-action and positive behaviour supports of the student with emotional and behavioural disorders. The qualitative study describes the teachers’ activities and experiences, defines the types of teachers’ activities and of developed interactions with the students with emotional and behavioural disorders. In order to go further into the developed interactions of the student with emotional and behavioural disorders, into the reality of the educational process and change it by encouraging evidence-based teachers’ activities and positive behaviour supports of the student with emotional and behavioural disorders, participatory action research was carried out. The reflection of the researcher’s role discloses the researcher’s versatile involvement and influence on the research process, permanently observing and actualising the researcher’s personal interaction with the participants, personal reactions, relating theoretical concepts, practical activities and reflections. The dissertation provides detailed definitions of the concepts of students with emotional and behavioural disorders, and of the structure and principles of evidence-based teacher’s activity; action took place in the reality of the... [to full text] / Disertacijoje pagrindžiamas įrodymais grįstos mokytojo veiklos veiksmingumas, ugdant emocijų ir elgesio sutrikimų turinčius mokinius bendrojo lavinimo mokykloje. Įrodymais grįstos mokytojo veiklos veiksmingumas analizuojamas dviem kriterijais: mokytojo mokymasis veikiant bei emocijų ir elgesio sutrikimų turinčio mokinio pozityvaus elgesio palaikymas. Atlikus kokybinį tyrimą, apibūdinamos mokytojų veiklos ir patirtys, apibrėžiami mokytojų vykdomų veiklų, konstruojamų sąveikų su emocijų ir elgesio sutrikimų turinčiu mokiniu tipai. Vykdyti veiklos tyrimai dalyvaujant, siekiant giliau pažinti emocijų ir elgesio sutrikimų turinčio mokinio ugdymo proceso realybę bei ją keisti, skatinant įrodymais grįstas mokytojų veiklas bei emocijų ir elgesio sutrikimų turinčio mokinio pozityvaus elgesio palaikymą. Atlikta tyrėjos vaidmens refleksija atskleidžia visapusišką tyrėjos įsitraukimą ir įtaką tyrimo procesui, nuolat stebint ir aktualizuojant asmeninę tyrėjos sąveiką su dalyviais, asmenines reakcijas, susiejant teorinius konceptus, praktines veiklas ir refleksijas. Disertacijoje detaliai apibrėžiami emocijų ir elgesio sutrikimų turinčių mokinių ugdymo konceptai, įrodymais grįstos mokytojo veiklos struktūra ir principai, pristatomos ugdymo proceso realybėje vykdytos veiklos. Remiantis atlikto disertacinio tyrimo rezultatais, pateikiamos empiriškai pagrįstos praktinės rekomendacijos mokiniams, tėvams, mokytojams ir specialistams bei mokslininkams.

This thesis reports on the development of a generic online system to support learning and teaching at the University of Melbourne. New online technologies, the fostering of innovation at national and university levels and my position within a central educational unit provided the opportunity in 1996 to adapt a previous software package for online use. My observations of the problematic nature of computer-facilitated learning (CFL) production led me to take an open approach to the development, seeking both a practical product and enhanced understanding. A series of formative questions defined the scope and goals of the study, which were to: *produce a generic online learning system; *increase understanding of the workplace experience of that development; and *develop an organisational model for the further development of generic CFL systems. Given this multi-disciplinary focus, many paradigms in the literature could potentially have guided the study. A number of these aligning with the research purposes, context and constructivist philosophy of the study, were reviewed from the perspectives of learning, CFL development and the organisation.

Made available in DSpace on 2015-04-16T14:49:19Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 968725 bytes, checksum: 4039440d7e53401862b617ddc7f82f33 (MD5) Previous issue date: 2010-01-20 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / The purpose of this study is to understand how the tax auditors from Renevue State Secretary of Paraíba learn in the professional practice context. With the growing demand for public services more comprehensive and effective, and the own engagement for a better profesionalization, to understand how effective their learning emerged as significant contribution to this process of professional development. To understand this field of study, it was fundamental perform studies in the field of learning organizational, specially those that emphasize contextualized professional learning, with emphasis on the role of experience and reflection. The research was guided by a qualitative approach within an interpretative perspective, since it will hear the professionals and know their stories in the context of fiscal actions, being the stories collected for the purposes of this research through the investigative method of Oral History. The study was conducted with ten tax auditors who work on Paraíba. The stories of the experiences and reflections revealed that the auditors learn initially and formally taught in the preparatory course, continuing his training in the context of practice, the interaction of senior auditors, the experience of collective professional practice, in the belonging in the auditors community of practice, the reflect on practices, individuals and collectives, the collective face of unexpected situations, revealing a process of a complex and dynamic learning. Also revealed that throughout his career, after the joining, playing "roles" in the "roadmap" of his professional learning, so it is possible to identify some learning facilitators inherent in this process. The fiscal practice ranges as approved by the known and unknown experience brought by new situations, having the reflection, individually and collectively, as final arbiter. Some new questions were left as references for future research, attention to the role of public service and its agents in the context of society. / O propósito deste estudo é compreender como os auditores fiscais da Secretaria de Estado da Receita da Paraíba aprendem no contexto da prática profissional. Com a crescente demanda por serviços públicos mais abrangentes e eficazes, e o próprio engajamento de seus quadros por melhor profissionalização, entender como se efetiva sua aprendizagem surge como relevante contribuição para esse processo de aperfeiçoamento profissional. Para entender esse campo de estudo, foi fundamental realizar estudos no campo da aprendizagem organizacional, notadamente aqueles que enfatizam a aprendizagem contextualizada na ação profissional, com destaque para o papel da experiência e da reflexão. A pesquisa foi orientada por uma abordagem qualitativa, dentro de uma perspectiva interpretativista, uma vez que tenciona ouvir os profissionais e saber suas histórias no contexto da ação fiscal, sendo essas histórias recolhidas para os fins desta pesquisa por meio do método investigativo da História Oral. O estudo foi realizado com dez auditores fiscais atuantes em postos fiscais paraibanos. As histórias das experiências e reflexões revelaram que os auditores aprendem inicialmente e formalmente no curso preparatório, prosseguindo sua formação no contexto da prática, no convívio com os auditores veteranos, na vivência de práticas profissionais coletivas, no pertencimento à comunidade de prática dos auditores, no refletir sobre as práticas, individuais e coletivas, no enfrentamento coletivo de situações inéditas, revelando um processo de aprendizagem complexo e dinâmico. Também revelaram que ao longo de sua trajetória, desde seu ingresso, desempenha papéis no roteiro de sua aprendizagem profissional, sendo possível identificar alguns mediadores da aprendizagem inerentes a esse processo. A prática fiscal oscila entre o conhecido referendado pela experiência e o desconhecido trazido pelas situações inéditas, tendo a reflexão, individual e coletiva, como árbitra final. Algumas novas questões foram deixadas como referências para futuras pesquisas, atentos para o papel do serviço público e seus agentes no contexto da sociedade.