This is the abstract book of the 3rd PhD conference at the TU Bergakademie Freiberg presenting the abstracts of the keynote speaker and the oral presentations.

“What is more real than the thoughts in your mind?”, “Re/structure your memories, re/construct your reality, re/define yourself”. The foundation of my research is about practising theory instead of theorising practice. My project begins with theory, which then leads to the science fiction film Mnemophrenia that constitutes the practical aspect of it. I attempt to demonstrate how theory and practice can be joined to create a fruitful union, each one feeding the other. In my research, I am inspired by Marshall McLuhan’s idea and use the medium as the message in order to depict and explore how cinema can affect human memory and more specifically create artificial memories and thus contribute to the dissolution of any boundaries between reality and fiction. The key research question that Mnemophrenia explores is: what would happen if in a future postmodern society the Bazinian myth of ‘total’ cinema becomes a reality? If ‘total cinema’ is pure realism and cinema can lead to artificial memories, then artificial memories and pure realism become one and films become artificial memories. Mnemophrenia depicts a different kind of human being or species, a schizophrenic ‘cyborg’ changed from within due to the advancement of virtual reality films which signals the end of cinema as we know it today. Mnemophrenia is about the future of cinema and maintains a horizon of hope that could lead to utopia; it does not discard technology as something evil as many previous science fiction films have done. I am interested in depicting through the film and examining in my thesis the possibility of a society where the dissolution of borders between fiction and reality does not lead to horrific consequences for humanity but instead promotes a potential for a new kind of identity that is an amalgam of real and artificial memories.

The majority of automatic speech recognition (ASR) systems rely on hidden Markov models (HMM), in which the output distribution associated with each state is modelled by a mixture of diagonal covariance Gaussians. Dynamic information is typically included by appending time-derivatives to feature vectors. This approach, whilst successful, makes the false assumption of framewise independence of the augmented feature vectors and ignores the spatial correlations in the parametrised speech signal. This dissertation seeks to address these shortcomings by exploring acoustic modelling for ASR with an application of a form of state-space model, the linear dynamic model (LDM). Rather than modelling individual frames of data, LDMs characterize entire segments of speech. An auto-regressive state evolution through a continuous space gives a Markovian model of the underlying dynamics, and spatial correlations between feature dimensions are absorbed into the structure of the observation process. LDMs have been applied to speech recognition before, however a smoothed Gauss-Markov form was used which ignored the potential for subspace modelling. The continuous dynamical state means that information is passed along the length of each segment. Furthermore, if the state is allowed to be continuous across segment boundaries, long range dependencies are built into the system and the assumption of independence of successive segments is loosened. The state provides an explicit model of temporal correlation which sets this approach apart from frame-based and some segment-based models where the ordering of the data is unimportant. The benefits of such a model are examined both within and between segments. LDMs are well suited to modelling smoothly varying, continuous, yet noisy trajectories such as found in measured articulatory data. Using speaker-dependent data from the MOCHA corpus, the performance of systems which model acoustic, articulatory, and combined acoustic-articulatory features are compared. As well as measured articulatory parameters, experiments use the output of neural networks trained to perform an articulatory inversion mapping. The speaker-independent TIMIT corpus provides the basis for larger scale acoustic-only experiments. Classification tasks provide an ideal means to compare modelling choices without the confounding influence of recognition search errors, and are used to explore issues such as choice of state dimension, front-end acoustic parametrization and parameter initialization. Recognition for segment models is typically more computationally expensive than for frame-based models. Unlike frame-level models, it is not always possible to share likelihood calculations for observation sequences which occur within hypothesized segments that have different start and end times. Furthermore, the Viterbi criterion is not necessarily applicable at the frame level. This work introduces a novel approach to decoding for segment models in the form of a stack decoder with A* search. Such a scheme allows flexibility in the choice of acoustic and language models since the Viterbi criterion is not integral to the search, and hypothesis generation is independent of the particular language model. Furthermore, the time-asynchronous ordering of the search means that only likely paths are extended, and so a minimum number of models are evaluated. The decoder is used to give full recognition results for feature-sets derived from the MOCHA and TIMIT corpora. Conventional train/test divisions and choice of language model are used so that results can be directly compared to those in other studies. The decoder is also used to implement Viterbi training, in which model parameters are alternately updated and then used to re-align the training data.

An intelligent agent must be capable of using its past experience to develop an understanding of how its actions affect the world in which it is situated. Given some objective, the agent must be able to effectively use its understanding of the world to produce a plan that is robust to the uncertainty present in the world. This thesis presents a novel computational framework called the Adaptive Modelling and Planning System (AMPS) that aims to meet these requirements for intelligence. The challenge of the agent is to use its experience in the world to generate a model. In problems with large state and action spaces, the agent can generalise from limited experience by grouping together similar states and actions, effectively partitioning the state and action spaces into finite sets of regions. This process is called abstraction. Several different abstraction approaches have been proposed in the literature, but the existing algorithms have many limitations. They generally only increase resolution, require a large amount of data before changing the abstraction, do not generalise over actions, and are computationally expensive. AMPS aims to solve these problems using a new kind of approach. AMPS splits and merges existing regions in its abstraction according to a set of heuristics. The system introduces splits using a mechanism related to supervised learning and is defined in a general way, allowing AMPS to leverage a wide variety of representations. The system merges existing regions when an analysis of the current plan indicates that doing so could be useful. Because several different regions may require revision at any given time, AMPS prioritises revision to best utilise whatever computational resources are available. Changes in the abstraction lead to changes in the model, requiring changes to the plan. AMPS prioritises the planning process, and when the agent has time, it replans in high-priority regions. This thesis demonstrates the flexibility and strength of this approach in learning intelligent behaviour from limited experience.

Analysing what it means being neither immigrant nor tourist in a foreign country, this thesis looks at the experiences of international PhD students and their supervisors in the Swedish university town Uppsala. The Faculty of Science and Technology of Uppsala University and the Swedish University of Agricultural Sciences (SLU) are the universities studied. Based on anthropological methods, this thesis focuses on the individual actors as they attempt to match the objective of doing a PhD while at the same time being in a different country. Using practice theory, the study highlights that the experiences of international PhD students and the people working with them vary greatly from PhD student to PhD student. Yet, almost all undergo a certain form of self-formation and adjustment process to match the requirements of not only their host country, but also the international field of science they attempt to enter. Being connected to other people in similar situations can help them adapt to this new environment, and many international PhD students join communities of practice. I argue that most develop an internationally recognised habitus, which helps them being recognised by the international community. The habitus can further, under certain circumstances, become partially conscious.

This research programme investigated the expression of prolyl hydroxylase (PHD) proteins in rodent tissues. The importance of PHD enzymes lies in their ability to render oxygen sensitivity to Hypoxia inducible factor (HIF), the principal mediator of intracellular oxygen homeostasis. The first part of this study focused on developing and validating anti-sera capable of detecting PHD proteins in rodent tissues. With these reagents, it was possible to assess the relative expression of each PHD protein in a number of different rat tissues. PHD2 was the most abundant isoform in all tissues studied. In contrast, an abundance of PHD1 was observed only in testis and skeletal muscle. A number of different tissue species of PHD3 were identified and their abundance was found to vary between different tissues. These observations provide further evidence of the principal role of PHD2 in regulating HIF in vivo, but also point towards additional roles for PHD1 and PHD3 in selected tissues. They highlight the potential for there being a complex interplay between different PHD enzymes which could, in the future, prove potential targets for therapeutic manipulation. This study also provides additional insights into the mechanisms underlying the phenotypes observed in PHD deletional mouse models which appear, in many cases, to be directly related to the abundance of a given PHD isoform. The emerging role of PHD3 as a promoter of sympathetic lineage apoptosis prompted further study of PHD3 expression in rat neuronal tissues. An abundance of PHD3 was demonstrated throughout the rat sympathetic nervous system, a finding which appeared at odds with its known role as a promoter of neuronal apoptosis and resulted in a series of collaborative studies which demonstrated a sympatho-adrenal phenotype in wild type compared to PHD3-/- mice. Further collaborative studies utilising wild type mice and those deleted of specific PHD isoforms, were carried out to assess the significance of the abundance of PHD3 and PHD1 noted here in rat hippocampus and testis respectively. While neither study demonstrated statistically significant phenotypes, these observations remain of interest and areas for future research.

This mixed-method study explored how students in a doctoral educational leadership cohort at one university used the cohort structure as a learning community or as a method of social support. Survey data were collected from 45 past and present cohort students and qualitative data were collected from three focus groups of 15 participants total. The survey measured four factors: General Cohort Experience, Trust Within the Cohort, Network, and Community of Learners. Quantitatively, one cohort was found to be significantly different from the others in terms of Trust Within the Cohort; and the 60 hour cohorts were found to be more satisfied with the cohort experience than the 48 hour doctoral cohorts at the university studied. The theme of trust and support from and to fellow members both during and after the cohort had dissolved was a strong recurrent theme in this study. Cohort members felt that they developed and strengthened their skills professionally as a result of participating in the cohort; however, they first had to develop a sense of community and trust with their fellow members in order to learn from them.

<div>The space around the Earth is becoming increasingly populated with a growth in number of launches and proliferation of debris. Currently, there are around 44,000 objects (with a minimum size of 10cm) orbiting the Earth as per the data made publicly available by the US strategy command (USSTRATCOM). These objects include active satellites and debris. The number of these objects are expected to increase rapidly in future from launches by companies in the private sector. For example, SpaceX is expected to deploy around 12000 new satellites in the LEO region to develop a space-based internet communication system. Hence in order to protect active space assets, tracking of all the objects is necessary. Probabilistic tracking methods have become increasingly popular for solving the multi-target tracking problem in Space Situational Awareness (SSA). This thesis studies one such technique known as the GM-PHD filter, which is an algorithm which estimates the number of objects and its states when non-perfect measurements (noisy measurements, false alarms) are available. For Earth orbiting objects, especially those in Geostationary orbits, ground based optical sensors are a cost-efficient way to gain information.In this case, the likelihood of gaining target-generated measurements depend on the probability of detection (p_D) of the target.An accurate modeling of this quantity is essential for an efficient performance of the filter. p_D significantly depends on the amount of light reflected by the target towards the observer. The reflected light depends on the relative position of the target with respect to the Sun and the observer, the shape, size and reflectivity of the object and the relative orientation of the object towards Sun and the observer. The estimation of the area and reflective properties of the object is in general, a difficult process. Uncontrolled objects, for example, start tumbling and no information regarding the attitude motion can be obtained. In addition, the shape can change because of disintegration and erosion of the materials. For the case of controlled objects, given that the object is stable, some information on the attitude can be obtained. But materials age in space which changes the reflective properties of the materials. Also, exact shape models for these objects are rare. Moreover,, area can never be estimated with optical measurements or any other measurements, as it is always albedo-area i.e., reflectivity times area that can be measured.</div><div> The purpose of this work is to design a variation of the GM-PHD filter which accounts for the uncertainty in p_D as the original GM-PHD filter designed by Vo and Ma assumes p_D as a constant. It is validated that the proposed method improves the filter performance when there is an uncertainty in area(hence uncertainty in p_D) of the targets. In the tested cases, the uncertainty in p_D was modeled as an uncertainty in area while assuming that the targets are spherical and that the reflectivity of the targets is constant. It is seen that a model mismatch in p_D affects the filter performance significantly and the proposed method improves the performance of the filter in all cases.</div>

Thesis (Ph.D.), Faculty of Humanities, School of Education, University of the Witwatersrand, 2009 / People who hold doctoral degrees are considered valuable national resources able to produce knowledge to address pressing problems, and important sources of labour for the higher education sector. However, in 2006, only 1100 people graduated with doctoral degrees in South Africa. This limits the potential for research and improvements in higher education. In addition, 618 of those graduates were white, making it difficult to address equity concerns. Within the higher education sector there are debates about how to increase enrolments in doctoral education and the best way to run PhD programmes for effective learning, high quality research results and for efficiency. But there is little South African-based empirical research into what makes people undertake PhDs, how the programmes work and what learning and knowledge result. This study explores how different stakeholders – national and institutional policymakers, academic staff and doctoral people – understand the PhD; how these understandings influence the practice of doctoral education; and how different practices affect the PhD experience and the learning and knowledge produced. The primary research question I address is: “How do existing models and pedagogies of doctoral programmes shape the learning of doctoral people and the outcomes of doctoral programmes in South Africa?” The origins of the Doctor of Philosophy degree are often traced back to the nineteenth century reforms of German universities when the idea emerged that all scholars should be actively involved in research. But this is a simplistic view. By examining the evolution of the PhD in greater depth, it becomes clear that it has undergone continuous change and has always served both the high-minded pursuit of knowledge and the more prosaic pursuit of skills for employment. The literature reflects ongoing tension between the scholarly view of the PhD as knowledge generation by an emerging scholar, and the labour market view of the PhD as developing high-level research skills. In the South African context both of these views can be observed, but I also identified a view of the PhD as ongoing personal development through an engagement with knowledge. The three views of the PhD are underpinned by different discourses which inform the practice of doctoral education. In South Africa, the traditional model of individual supervision dominates, and it varies by discipline, department and supervisor. But patterns of practice can be discerned and I identify four of these and discuss how supervisors construct their individual supervision practice. Doctoral education is also a function of the people who do PhDs. Much of the research undertaken in the overdeveloped world focuses on younger people who are starting out on academic careers. However, in South Africa, many people doing PhDs are older and midway through careers which are often not academic. This leads me to propose a model of intersecting contexts, as an alternative to McAlpine and Norton‟s nested context model of doctoral education, which more accurately reflects the local situation. I discuss the PhD experience and make use of the intersecting contexts model to develop the notion of congruence between the PhD, the contexts and the PhD person with more positive experiences being related to higher degrees of congruence. Finally, I consider how the outcomes of doctoral education, the learning and knowledge which result, relate to the expectations of the different stakeholders. The research took the form of a qualitative study with a multiple case-study design employing theoretical replication. I examined doctoral education in four academic units at three South African universities with the units selected to represent different disciplines. All four units were in previously advantaged universities from the English-speaking tradition and all were successfully producing PhD graduates. These rich pictures of how doctoral education takes place contribute empirical evidence to current debates about the PhD in South Africa. At a conceptual level I identify the competing discourses about what a PhD is. I provide a more nuanced understanding of the practice of doctoral education within the overarching model of individual supervision. The intersecting contexts model provides a way to understand the expectations and circumstances of doctoral people and the notion of congruence illuminates their varied experiences. Finally, the study confirms that the outcomes of doctoral education, in terms of learning and knowledge generated, meet at least some of the expectations of policy-makers, supervisors and people who do PhDs.

The development of the next generation of researchers is a priority if South Africa is to make a significant contribution to the international knowledge economy and establish itself as a force to be reckoned with in international research circles. In the context of this knowledge economy, researchers are increasingly being recognised as agents of economic growth. In order to be competitive, therefore, an extensive pool of active researchers needs to be cultivated. One way of doing this is to promote and develop doctoral capacity at the country’s universities. This entails, among other things, a move away from the traditional focus on what the supervisor does, to a more student-centred understanding of how the doctoral candidate experiences the process, and by implication, how this impacts on their research contribution. In this qualitative study, thirty doctoral candidates from the Faculties of Science and Humanities at the University of the Witwatersrand, Johannesburg, were interviewed in order to establish firstly, how they had experienced their supervision at this level, and secondly, whether or not these experiences had influenced the successful completion of their doctorates. One of the principal findings of the research was that the role of the supervisor was not central to the achievement of their degree; rather, many of the doctoral candidates asserted a significant level of agency in both identifying and making contact with experts beyond their university-appointed supervisors in order to supplement their access to relevant knowledge. In the majority of cases, this was encouraged by the supervisors. Such enterprises represent a marked departure from the traditional models of supervision in the Science and Humanities faculties. In the case of the former, the customary co-supervision arrangement is increasingly being augmented by student-initiated collaboration with authorities located outside the formal boundaries of the institution. The traditional Humanities model of supervision is also transforming from a one-on-one relationship to a style characterised by multiple supervisors, each from separate but cognate disciplines. The research suggested that these emergent models are eclipsing their predecessors as doctoral candidates increasingly recognise the value of extending the breadth of their disciplinary exposure beyond the confines of the university.