Middle childhood is a critical period of attentional development. Previous research has linked neural oscillations in the theta frequency band to controlled attentional and cognitive processes, which has been replicated in children and adults. The development of executive attention, which biases attention and alters mental representation in the service of task goals, is preceded by development of sustained attention, and further selective sustained attention. These three attentional constructs can be represented by Posner’s altering (sustained) orienting (selective sustained) and executive attention networks. Effortful control, a temperament trait describing individual differences in ability to exert self-regulation, has been linked to efficiency of the executive attention system. To examine attentional engagement (within task) and demand (between task) electroencephalography was recorded from 226 six- and nine-year-old children at medial and lateral, frontal, and parietal scalp locations during a baseline, visual search, and the Attention Network Task to measure sustained, selective sustained and executive attention, respectively. Repeated measures MANOVA of frontal and parietal scalp locations indicate multiple complex three-way interactions of region (medial vs lateral), Age, and Block/Task. Frontal and parietal activation patterns were also different from each other, as well as between age groups. When temperament factors, effortful control and surgency, were included in the model (repeated measures MANCOVA) most interactions were no longer significant. We therefore find, in accord with previous literature, that medial frontal theta is impacted by attentional engagement and demand but this association is heavily impacted by individual biologically based differences. / M.S. / During middle childhood, kids' ability to pay attention develops into a more sophisticated, adult-like form. Scientists have found that the way our brain waves work in a certain frequency (called theta) is connected to our ability to focus and think. This is true for both kids and adults. There are three critical forms of attention identified by developmental and cognitive researchers. First, there's the kind where you can stay focused on something for a while. Then, there's another type where you not only stay focused but also pick out specific things to focus on. Lastly, there's the kind where you can change your focus to fit the task you're doing. Our goal was to examine how theta brain waves relate to each of these forms of attention and how those change after three years. Using the electroencephalography technique, we measured brain activity of used a special brain scanning technique on 226 kids when they were six and nine years old, while they completed three tasks. One analysis focused on attentional engagement, how children focused over the course of a single task, and the other on attentional demand, how children focused differently as tasks got more difficult. We found power in the theta frequency band decreased with age, which means that children’s attentional processing was more efficient the older they were. We also found that theta in the front of the brain did not change greatly over the course of the task except for the initial set of trials. This was different from the middle regions of the brain, which changed a lot over the course of the task. Theta power in both frontal and middle parts of the brain was different between the tasks, and harder tasks were associated with more theta. Finally, we found that temperament, a child’s individual self-control and excitability, greatly explained the differences in theta power over the tasks.

During the last few years, segmentation architectures based on deep learning achieved promising results. On the other hand, attention networks have been invented years back and used in different tasks but rarely used in medical applications. This thesis investigated four main attention mechanisms; Squeeze and Excitation, Dual Attention Network, Pyramid Attention Network, and Attention UNet to be used in medical image segmentation. Also, different hybrid architectures proposed by the author were tested. Methods were tested on a kidney tumor dataset and against UNet architecture as a baseline. One version of Squeeze and Excitation attention outperformed the baseline. Original Dual Attention Network and Pyramid Attention Network showed very poor performance, especially for the tumor class. Attention UNet architecture achieved close results to the baseline but not better. Two more hybrid architectures achieved better results than the baseline. The first is a modified version of Squeeze and Excitation attention. The second is a combination between Dual Attention Networks and UNet architecture. Proposed architectures outperformed the baseline by up to 3% in tumor Dice coefficient. The thesis also shows the difference between 2D architectures and their 3D counterparts. 3D architectures achieved more than 10% higher tumor Dice coefficient than 2D architectures.

abstract: Virtual digital assistants are automated software systems which assist humans by understanding natural languages such as English, either in voice or textual form. In recent times, a lot of digital applications have shifted towards providing a user experience using natural language interface. The change is brought up by the degree of ease with which the virtual digital assistants such as Google Assistant and Amazon Alexa can be integrated into your application. These assistants make use of a Natural Language Understanding (NLU) system which acts as an interface to translate unstructured natural language data into a structured form. Such an NLU system uses an intent finding algorithm which gives a high-level idea or meaning of a user query, termed as intent classification. The intent classification step identifies the action(s) that a user wants the assistant to perform. The intent classification step is followed by an entity recognition step in which the entities in the utterance are identified on which the intended action is performed. This step can be viewed as a sequence labeling task which maps an input word sequence into a corresponding sequence of slot labels. This step is also termed as slot filling. In this thesis, we improve the intent classification and slot filling in the virtual voice agents by automatic data augmentation. Spoken Language Understanding systems face the issue of data sparsity. The reason behind this is that it is hard for a human-created training sample to represent all the patterns in the language. Due to the lack of relevant data, deep learning methods are unable to generalize the Spoken Language Understanding model. This thesis expounds a way to overcome the issue of data sparsity in deep learning approaches on Spoken Language Understanding tasks. Here we have described the limitations in the current intent classifiers and how the proposed algorithm uses existing knowledge bases to overcome those limitations. The method helps in creating a more robust intent classifier and slot filling system. / Dissertation/Thesis / Masters Thesis Computer Science 2018

The prefrontal cortices, in particular lateral prefrontal cortex (LPFC) and anterior cingulate cortex (ACC), have been implicated in top-down control of attention switching and behavioral flexibility. These cortices and their networks are disrupted in autism, a condition in which diverse behaviors such as social communication and attention control are dysregulated. However, little is known about the typical development of these cortical areas or the ways in which this process is altered in neurodevelopmental disorders. In order to identify changes that could affect the local processing of signals transmitted by the short-range pathways connecting the ACC and LPFC I assessed developmental changes in the distinct cortical layers, which send and receive different pathways and have unique inhibitory microenvironments that dictate excitatory-inhibitory balance. Normative developmental trends were compared with those seen in individuals with autism to identify changes that may contribute to symptoms of attention dysfunction. Unbiased quantitative methods were used to study overall neuron density, the density of inhibitory neurons labeled by the calcium-binding proteins calbindin (CB), calretinin (CR), and parvalbumin (PV), and the density, size, and trajectory of myelinated axons in the individual cortical layers in children and adults with and without a diagnosis of autism. There was a reduction in neuron density and an increase in the density of myelinated axons in both areas during neurotypical development. Axons in layers 1-3 of LPFC were disorganized in autism, with increased variability in the trajectory of axons in children and a decrease in the proportion of thin axons in adults. These findings were most significant in layer 1, the ultimate feedback-receiving layer in the cortex. While there were no differences in neuron populations between cohorts in children, in adults with autism there was a significant reduction in the density of CR-expressing neurons in LPFC layers 2-6 and a significant increase in the density of PV-expressing neurons in ACC layers 5-6. In autism, these findings suggest that dysregulation of the normal development of axonal networks, seen in children, may induce compensatory developmental changes in cell and axon populations in adults that could be connected to attention dysregulation. / 2021-10-07T00:00:00Z

No description available.

Dialogues are an interesting type of document, as they contain a speaker role feature not found in other types of texts. Previous work has included incorporating a speaker role dependency in text-generation, but little has been done in the realm of text classification. In this thesis, we incorporate speaker role dependency in a classification model by creating different speaker dependent word representations and simulating a conversation within neural networks. The results show a significant improvement in the performance of the binary classification of dialogues, with incorporated speaker role information. Further, by extracting attention weights from the model, we are given an insight into how the speaker’s role affects the interpretation of utterances, giving an intuitive explanation of our model. / Konversationer är en speciell typ av text, då den innehåller information om talare som inte hittas i andra typer av dokument. Tidigare arbeten har inkluderat en talares roll i generering av text, men lite har gjorts inom textklassificering. I det här arbetet, introducerar vi deltagarens roller till en klassifikationsmodell. Detta görs genom att skapa ordrepresentationer, som är beroende på deltagaren i konversationen, samt simulering av en konversation inom ett neuralt nätverk. Resultaten visar en signifikant förbättring av prestandan i binär klassificering av dialoger, med talares roll inkluderat. Vidare, genom utdragning av attentionvikterna, kan vi få en bättre överblick över hur en talares roll påverkar tolkningen av yttranden, vilket i sin tur ger en mer intuitiv förklaring av vår modell.