Mineração de fluxos contínuos de dados é uma área de pesquisa emergente que visa extrair conhecimento a partir de grandes quantidades de dados, gerados continuamente. Detecção de novidade é uma tarefa de classificação que consiste em reconhecer que um exemplo ou conjunto de exemplos em um fluxo de dados diferem significativamente dos exemplos vistos anteriormente. Essa é uma importante tarefa para fluxos contínuos de dados, principalmente porque novos conceitos podem aparecer, desaparecer ou evoluir ao longo do tempo. A maioria dos trabalhos da literatura apresentam a detecção de novidade como uma tarefa de classificação binária. Poucos trabalhos tratam essa tarefa como multiclasse, mas usam medidas de avaliação binária. Em vários problemas, o correto seria tratar a detecção de novidade em fluxos contínuos de dados como uma tarefa multiclasse, no qual o conceito conhecido do problema é formado por uma ou mais classes, e diferentes novas classes podem aparecer ao longo do tempo. Esta tese propõe um novo algoritmo MINAS para detecção de novidade em fluxos contínuos de dados. MINAS considera que a detecção de novidade é uma tarefa multiclasse. Na fase de treinamento, MINAS constrói um modelo de decisão com base em um conjunto de exemplos rotulados. Na fase de aplicação, novos exemplos são classificados usando o modelo de decisão atual, ou marcados como desconhecidos. Grupos de exemplos desconhecidos podem formar padrões-novidade válidos, que são então adicionados ao modelo de decisão. O modelo de decisão é atualizado ao longo do fluxo a fim de refletir mudanças nas classes conhecidas e permitir inserção de padrões-novidade. Esta tese também propõe uma nova metodologia para avaliação de algoritmos para detecção de novidade em fluxos contínuos de dados. Essa metodologia associa os padrões-novidade não rotulados às classes reais do problema, permitindo assim avaliar a matriz de confusão que é incremental e retangular. Além disso, a metodologia de avaliação propõe avaliar os exemplos desconhecidos separadamente e utilizar medidas de avaliação multiclasse. Por último, esta tese apresenta uma série de experimentos executados usando o MINAS e os principais algoritmos da literatura em bases de dados artificiais e reais. Além disso, o MINAS foi aplicado a um problema real, que consiste no reconhecimento de atividades humanas usando dados de acelerômetro. Os resultados experimentais mostram o potencial do algoritmo e da metodologia propostos / Data stream mining is an emergent research area that aims to extract knowledge from large amounts of continuously generated data. Novelty detection is a classification task that assesses if an example or a set of examples differ significantly from the previously seen examples. This is an important task for data streams, mainly because new concepts may appear, disappear or evolve over time. Most of the work found in the novelty detection literature presents novelty detection as a binary classification task. A few authors treat this task as multiclass, but even they use binary evaluation measures. In several real problems, novelty detection in data streams must be treated as a multiclass task, in which, the known concept about the problem is composed by one or more classes and different new classes may appear over time. This thesis proposes a new algorithm MINAS for novelty detection in data streams. MINAS deals with novelty detection as a multiclass task. In the training phase, MINAS builds a decision model based on a labeled data set. In the application phase, new examples are classified using the decision model, or marked with an unknown profile. Groups of unknown examples can be later used to create valid novelty patterns, which are added to the current decision model. The decision model is updated as new data arrives in the stream in order to reflect changes in the known classes and to allow the addition of novelty patterns. This thesis also proposes a new methodology to evaluate classifiers for novelty detection in data streams. This methodology associates the unlabeled novelty patterns to the true problem classes, allowing the evaluation of a confusion matrix that is incremental and rectangular. In addition, the proposed methodology allows the evaluation of unknown examples separately and the use multiclass evaluation measures. Additionally, this thesis presents a set of experiments carried out comparing the MINAS algorithm and the main novelty detection algorithms found in the literature, using artificial and real data sets. Finally, MINAS was applied to a human activity recognition problem using accelerometer data. The experimental results show the potential of the proposed algorithm and methodologies

Novel sounds in the auditory oddball paradigm elicit a biphasic dilation of the pupil (PDR) and P3a as well as novelty P3 event-related potentials (ERPs). The biphasic PDR has been hypothesized to reflect the relaxation of the iris sphincter muscle due to parasympathetic inhibition and the constriction of the iris dilator muscle due to sympathetic activation. We measured the PDR and the P3 to neutral and to emotionally arousing negative novels in dark and moderate lighting conditions. By means of principal component analysis (PCA) of the PDR data we extracted two components: the early one was absent in darkness and, thus, presumably reflects parasympathetic inhibition, whereas the late component occurred in darkness and light and presumably reflects sympathetic activation. Importantly, only this sympathetic late component was enhanced for emotionally arousing (as compared to neutral) sounds supporting the hypothesis that emotional arousal specifically activates the sympathetic nervous system. In the ERPs we observed P3a and novelty P3 in response to novel sounds. Both components were enhanced for emotionally arousing (as compared to neutral) novels. Our results demonstrate that sympathetic and parasympathetic contributions to the PDR can be separated and link emotional arousal to sympathetic nervous system activation.

Condition-based maintenance is an important aspect in various industries to ensure reliable operation of machinery. To successfully execute maintenance responsibilities, it is required to know which components are healthy and which are in a damaged state. Thus, the need for effective incipient fault detection requires a method that can separate fault signatures from operating condition information. Conventional gearbox monitoring techniques assume that a change in the vibration signal is caused by the presence of a fault. Under constant operating conditions this assumption may be valid, but under fluctuating conditions the assumption does not hold. Fluctuating operating conditions are inevitable for gearboxes in mining and wind turbine industries due to fluctuating ground and wind properties. The fluctuating conditions cause smearing of the signal frequency spectrum and valuable diagnostic information is lost when using classical condition monitoring techniques. More sophisticated signal processing techniques are therefore needed to effectively diagnose incipient faults to make informed asset management decisions. In this dissertation, envelope analysis, which has long been recognized as one of the best methods for bearing fault diagnosis, is used as the primary diagnostic tool. A common precursor to envelope analysis is bandpass filtering which is aimed at emphasising bearing faults and removing background noise and deterministic components. Identification and optimal selection of the informative frequency band which contains damage related information is the focus area for research in this dissertation. Many automatic band selection techniques exist and have proven effective under constant speed conditions. However, it has been shown that these techniques occasionally identify frequency bands that contain non-damage related information, especially under fluctuating speeds and low damage levels. With this research, a new methodology is proposed which makes use of popular informative frequency band selection techniques, such as the Fast Kurtogram amongst others, to effectively identify damage under constant and fluctuating speed conditions. The proposed methodology uses both healthy and damaged vibration signals to identify novelty information. In doing so, the method can also identify damage earlier than existing methods. The technique is designed to ignore potentially dominant deterministic components which would lead to incorrect band selection for envelope analysis. Furthermore, pre-whitening of vibration signals is a common technique to enhance the bearing signal-to-noise ratio. Without pre-whitening, random noise and deterministic components often dominate the bearing fault signatures and render existing diagnostic techniques ineffective. The proposed methodology is shown to be more robust than existing automatic band selection methods because it requires no pre-whitening. By using both healthy and damaged signals, the proposed methodology favours frequency bands that contain damage related information. The findings in this dissertation are validated on a range of synthetic signals as well as on actual experimental data. The synthetic signals are constructed from a phenomenological gearbox model where the exact operating and bearing condition can be controlled. The experimental results are statistically compared for a wide range of signals and damage levels such that the robustness of the proposed method can be critically evaluated. It was found that the new method is capable of outperforming existing methods in terms of percentage classification of bearing signals with outer race damage and can detect damage with smaller fault severity. / Dissertation (MEng)--University of Pretoria, 2019. / Mechanical and Aeronautical Engineering / MEng / Unrestricted

This thesis examines innovation and design processes in small established companies. There is a great interest in this area yet paradoxically the area is under-researched, since most innovation research is done on large companies. The research questions are: How do small established companies carry out their innovation and design processes? and How does the context and novelty of the process and product affect the same processes? The thesis is built on three research papers that used the research method of multiple case studies of different small established companies. The innovation and design processes found were highly context dependent and were facilitated by committed resources, a creative climate, vision, low family involvement, delegated power and authority, and linkages to external actors such as customers and users. Both experimental cyclical and linear structured design processes were found. The choice of structure is explained by the relative product and process novelty experienced by those developing the product innovation. Linear design processes worked within a low relative novelty situation and cyclical design processes worked no matter the relative novelty. The innovation and design processes found were informal, with a low usage of formal systematic design methods, except in the case of design processes for software. The use of formal systematic methods in small companies seems not always to be efficient, because many of the problems the methods are designed to solve are not present. Customers and users were found to play a large and important role in the innovation and design processes found and gave continuous feedback during the design processes. Innovation processes were found to be intertwined, yielding synergy effects, but it was common that resources were taken from the innovation processes for acute problems that threatened the cash flow. In sum, small established companies have the natural prerequisites to take advantage of lead-user inventions and cyclical design processes. Scarce resources were found to be the main factor hindering innovation, but the examined companies practiced several approaches to increase their resources or use existing scarce resources more efficiently in their innovation and design processes. Examples of these approaches include adopting lead-user inventions and reducing formality in the innovation and design processes.

Our attentional system has the remarkable ability to allow familiar contexts to guide attentional orienting, while still retaining the ability to orient rapidly to novelty in our environment. Many cognitive paradigms have been used to investigate the particular process that is responsible for orienting attention to novel events, but each paradigm has produced a unique set of boundary conditions. One such paradigm has studied an effect labelled Inhibition of Return (IOR), which has been argued to tap into an attentional mechanism that rapidly orients attention to novelty, but only in the spatial domain. The IOR effect was initially taken as evidence of a fundamental difference between spatial attentional orienting and non-spatial attentional orienting. However, there were a small number of early studies that questioned the view that the IOR effect can only be observed in the spatial domain. In this dissertation, I built upon the evidence for non-spatial IOR by uncovering the effect using a Target-Target (TT) procedure. Although a number of prior studies had failed to observe non-spatial IOR using a TT procedure, I was able to uncover non-spatial IOR effects using a TT procedure by introducing an intervening event. The IOR-like effect that was uncovered using this procedure was labelled the intervening event effect. I introduced a dual process framework to explain the intervening event effect. According to the dual process framework, intervening events between consecutive targets can disrupt an episodic integration process, allowing the influence of a separate opposing process to be measured more directly. Using the dual process framework, I studied the level of processing of the intervening event that was necessary to disrupt episodic integration, as well as the context-sensitivity of the episodic integration process. Lastly, I investigated the role of subjective expectancy in the studies used to measure non-spatial IOR in this thesis. / Thesis / Doctor of Philosophy (PhD)

Across many scientific disciplines, there has emerged an open opportunity to utilize the scale and reach of the Internet to collect scientific contributions from scientists and non-scientists alike. This process, called citizen science, has already shown great promise in the fields of biology and astronomy. Within the fields of artificial life (ALife) and evolutionary computation (EC) experiments in collaborative interactive evolution (CIE) have demonstrated the ability to collect thousands of experimental contributions from hundreds of users across the glob. However, such collaborative evolutionary systems can take nearly a year to build with a small team of researchers. This dissertation introduces a new developer framework enabling researchers to easily build fully persistent online collaborative experiments around almost any evolutionary domain, thereby reducing the time to create such systems to weeks for a single researcher. To add collaborative functionality to any potential domain, this framework, called Worldwide Infrastructure for Neuroevolution (WIN), exploits an important unifying principle among all evolutionary algorithms: regardless of the overall methods and parameters of the evolutionary experiment, every individual created has an explicit parent-child relationship, wherein one individual is considered the direct descendant of another. This principle alone is enough to capture and preserve the relationships and results for a wide variety of evolutionary experiments, while allowing multiple human users to meaningfully contribute. The WIN framework is first validated through two experimental domains, image evolution and a new two-dimensional virtual creature domain, Indirectly Encoded SodaRace (IESoR), that is shown to produce a visually diverse variety of ambulatory creatures. Finally, an Android application built with WIN, "filters, allows users to interactively evolve custom image effects to apply to personalized photographs, thereby introducing the first CIE application available for any mobile device. Together, these collaborative experiments and new mobile application establish a comprehensive new platform for evolutionary computation that can change how researchers design and conduct citizen science online.

No description available.

No description available.

No description available.

No description available.