Thèses sur le sujet « Mechanism of attention »
Pour voir les autres types de publications sur ce sujet consultez le lien suivant : Mechanism of attention.
Créez une référence correcte selon les styles APA, MLA, Chicago, Harvard et plusieurs autres
Consultez les 50 meilleures thèses pour votre recherche sur le sujet « Mechanism of attention ».
À côté de chaque source dans la liste de références il y a un bouton « Ajouter à la bibliographie ». Cliquez sur ce bouton, et nous générerons automatiquement la référence bibliographique pour la source choisie selon votre style de citation préféré : APA, MLA, Harvard, Vancouver, Chicago, etc.
Vous pouvez aussi télécharger le texte intégral de la publication scolaire au format pdf et consulter son résumé en ligne lorsque ces informations sont inclues dans les métadonnées.
Parcourez les thèses sur diverses disciplines et organisez correctement votre bibliographie.
1
Fitzgerald, Marilyn. « Are attention bias and interpretation bias reflections of a single common mechanism or multiple independent mechanisms ? » University of Western Australia. School of Psychology, 2008. http://theses.library.uwa.edu.au/adt-WU2009.0052.
Texte intégralRésumé :
There is abundant evidence of anxiety-linked threat-biased attention and anxiety-linked threat-biased interpretation (cf. Mathews & MacLeod, 1994, 2005). The present research aimed to determine whether these cognitive biases reflect a single common underlying mechanism (the Common Mechanism Account) or multiple independent underlying mechanisms (the Independent Mechanisms Account). To address this question, a battery of eight experimental tasks was developed; four tasks measured attention bias and four measured interpretation bias. Participants with different levels of trait anxiety, completed pairs of these tasks. The pattern of associations amongst all eight tasks was compared with the pattern of associations between the four tasks that measured attention bias and the pattern of associations between the four tasks that measured interpretation bias. Both Accounts predicted strong associations between the four tasks that measured attention bias, and between the four tasks that measured interpretation bias. However, the Common Mechanism Account predicted generally strong associations between all of the eight tasks, that were equivalent in strength to the associations between tasks measuring attention bias and to the associations between tasks measuring interpretation bias. In contrast, the Independent Mechanisms Account predicted weaker associations between all of the eight tasks than the associations either between the tasks measuring attention bias or between the tasks measuring interpretation bias. The obtained pattern of associations between internally reliable measures of anxiety-linked attention bias and anxiety-linked interpretation bias failed to support the Common Mechanism Account, but rather was consistent with the predictions of the Independent Mechanisms Account. Theoretical and applied implications of the results are discussed.
2
Yan, Shiyang. « Visual attention mechanism in deep learning and its applications ». Thesis, University of Liverpool, 2018. http://livrepository.liverpool.ac.uk/3028892/.
Texte intégralRésumé :
Recently, in computer vision, a branch of machine learning, called deep learning, has attracted high attention due to its superior performance in various computer vision tasks such as image classification, object detection, semantic segmentation, action recognition and image description generation. Deep learning aims at discovering multiple levels of distributed representations, which have been validated to be discriminatively powerful in many tasks. Visual attention is an ability of the vision system to selectively focus on the salient and relevant features in a visual scene. The core objective of visual attention is to achieve the least possible amount of visual information to be processed to solve the complex high-level tasks, e.g., object recognition, which can lead the whole vision process to become effective. The visual attention is not a new topic which has been addressed in the conventional computer vision algorithms for many years. The development and deployment of visual attention in deep learning algorithms are of vital importance since the visual attention mechanism matches well with the human visual system and also shows an improving effect in many real-world applications. This thesis is on the visual attention in deep learning, starting from the recent progress in visual attention mechanism, followed by several contributions on the visual attention mechanism targeting at diverse applications in computer vision, which include the action recognition from still images, action recognition from videos and image description generation. Firstly, the soft attention mechanism, which was initially proposed to combine with Recurrent Neural Networks (RNNs), especially the Long Short-term Memories (LSTMs), was applied in image description generation. In this thesis, instead, as one contribution to the visual attention mechanism, the soft attention mechanism is proposed to directly plug into the convolutional neural networks for the task of action recognition from still images. Specifically, a multi-branch attention network is proposed to capture the object that the human is intereating with and the scene in which the action is performing. The soft attention mechanism applying in this task plays a significant role in capturing multi-type contextual information during recognition. Also, the proposed model can be applied in two experimental settings: with and without the bounding box of the person. The experimental results show that the proposed networks achieved state-of-the-art performance on several benchmark datasets. For the action recognition from videos, our contribution is twofold: firstly, the hard attention mechanism, which selects a single part of features during recognition, is essentially a discrete unit in a neural network. This hard attention mechanism shows superior capacity in discriminating the critical information/features for the task of action recognition from videos, but is often with high variance during training, as it employs the REINFORCE algorithm as its gradient estimator. Hence, this brought another critical research question, i.e., the gradient estimation of the discrete unit in a neural network. In this thesis, a Gumbel-softmax gradient estimator is applied to achieve this goal, with much lower variance and more stable training. Secondly, to learn a hierarchical and multi-scale structure for the multi-layer RNN model, we embed discrete gates to control the information between each layer of the RNNs. To make the model differentiable, instead of using the REINFORCE-like algorithm, we propose to use Gumbel-sigmoid to estimate the gradient of these discrete gates. For the task of image captioning, there are two main contributions in this thesis: primarily, the visual attention mechanism can not only be used to reason on the global image features but also plays a vital role in the selection of relevant features from the fine-grained objects appear in the image. To form a more comprehensive image representation, as a contribution to the encoder network for image captioning, a new hierarchical attention network is proposed to fuse the global image and local object features through the construction of a hierarchical attention structure, to better the visual representation for the image captioning. Secondly, to solve an inherent problem called exposure-biased issue of the RNN-based language decoder commonly used in image captioning, instead of only relying on the supervised training scheme, an adversarial training-based policy gradient optimisation algorithm is proposed to train the networks for image captioning, with improved results on the evaluation metrics. In conclusion, comprehensive research has been carried out for the visual attention mechanism in deep learning and its applications, which include action recognition and image description generation. Related research topics have also been discussed, for example, the gradient estimation of the discrete units and the solution to the exposure-biased issue in the RNN-based language decoder. For the action recognition and image captioning, this thesis presents several contributions which proved to be effective in improving existing methods.
3
Parker, Amanda Louise. « A cross-modal investigation into the relationships between bistable perception and a global temporal mechanism ». Thesis, The University of Sydney, 2013. http://hdl.handle.net/2123/9545.
Texte intégralRésumé :
When the two eyes are presented with sufficiently different images, Binocular Rivalry (BR) occurs. BR is a form of bistable perception involving stochastic alternations in awareness between distinct images shown to each eye. It has been suggested that the dynamics of BR are due to the activity of a central temporal process and are linked to involuntary mechanisms of selective attention (aka exogenous attention). To test these ideas, stimuli designed to evoke exogenous attention and central temporal processes were employed during BR observation. These stimuli included auditory and visual looming motion and streams of transient events of varied temporal rate and pattern. Although these stimuli exerted a strong impact over some aspects of BR, they were unable to override its characteristic stochastic pattern of alternations completely. It is concluded that BR is subject to distributed influences, but ultimately, is achieved in neural processing areas specific to the binocular conflict.
4
Raykos, Bronwyn C. « Attentional and interpretive biases : independent dimensions of individual difference or expressions of a common selective processing mechanism ? » University of Western Australia. School of Psychology, 2007. http://theses.library.uwa.edu.au/adt-WU2007.0018.
Texte intégralRésumé :
[Truncated abstract] Attentional and interpretive biases are important dimensions of individual difference that have been implicated in the etiology and maintenance of a range of clinical problems. Yet there has been no systematic investigation into the relationship between these dimensions of individual difference. The current research program tested predictions derived from two competing theoretical accounts of the relationship between attentional and interpretive biases. The Common Mechanism Account proposes that cognitive biases represent concurrent manifestations of a single underlying selective processing mechanism. The Independent Mechanism account proposes that independent mechanisms underlie each bias. . . An apparent contradiction is that the manipulation of one bias served to also modify the other bias, despite the observation that the magnitude of the resulting change in both biases was uncorrelated. Neither the Common Mechanism nor the Independent Pathways accounts can adequately explain this pattern of results. A new account is proposed, in which attentional and interpretive biases are viewed as representing mechanisms that are related but that are not the same. Theoretical and applied implications of these findings are discussed, including the possibility that the two biases each may best predict emotional reactions to quite different stressful events and that training programs designed to attenuate allocation of attentional resources to threat may serve to reduce both attentional and interpretive selectivity in emotionally vulnerable individuals.
5
Wang, Jing. « Hyperspectral Image Classification Based on Deep Learning and Module Inspired by Human Attention Mechanism ». Thesis, Griffith University, 2020. http://hdl.handle.net/10072/397634.
Texte intégralRésumé :
Hyperspectral imaging technology acquires image data in a number of continuous narrow bands of the electromagnetic wave. The obtained hyperspectral images contain details of spectral re ectance of targets in addition to spatial information. The ability to characterize abundant spectral details of hyperspectral image makes it particularly suitable for remote sensing image analysis. Hyperspectral remote sensing image classi cation is one of the most important applications in remote sensing, and is the main research problem of this thesis.
Researchers have already proposed a large variety of methods for hyperspectral image classi cation in the last few decades, which can be categorized into traditional methods and deep learning based methods. Recently, with the development of high performance computing and collection of large datasets, deep learning methods have been state of the art in hyperspectral image classi cation. Most of the existing deep learning methods take in the hyperspectral image and learn discriminant features in plain convolutional or fully connected layers. This learning manner treats all raw pixels and extracted features equally. However, human brains do not perform recognition task with equal consideration of every involved element. For recognition or classi cation tasks, it is possible that some parts of inputs or features are more important, while others are useless. Our visual system has the capability of attending to the signi cant aspects and ignoring irrelevant components. This has greatly contributed to our cognition ability and e ciency. Inspired by the attention mechanism of human brain, we design corresponding attention modules in the context of arti cial neural network for hyperspectral image classi cation. In addition, human visual system is a universal feature extractor and classi er in the sense that we can perform classi cation across multiple image styles, modalities and distributions. On the contrary, current deep learning based hyperspectral classi - cation paradigms require an individual model for every data domain. This is expensive and ine cient. Following similar philosophy of attention mechanism, we design domain attention modules for multi-domain hyperspectral image classi cation.
In this thesis, we propose three attention modules for deep learning based hyperspectral image classi cation. In the rst work, we introduce attention based feature weighting networks for improving the classi cation accuracy of current plain neural networks. In a deep network for hyperspectral application, a hierarchy of spectral or spatial features are extracted layer by layer. Each layer contains the same semantic level of features. To model the importance of features in the same level, attention modules are designed by branching from current feature maps. In the attention branch, three steps are executed: summarizing information from current layer, modeling relationship among the features with fully connected or convolution layers, and outputting weighting masks to be multiplied with the original features. We propose feature weighting attention modules for spectral CNN, spatial CNN and spectral-spatial CNN, respectively.
In the second work, we design attention modules speci cally attending to the bands of hyperspectral image. Compared to hidden features extracted in hidden layers of neural networks which have less interpretability and physical meaning, spectral bands of hyperspectral images correspond directly to real wavelength in the physical world. Thus attending to bands has special importance in a couple of aspects. First, it in uences the design and cost of hyperspectral sensor. Second, it is directly related to the dimension of the obtained raw data. Our band attention module can perform both band weighting and band selection. For band weighting, it has the ability to assign sample-wise weights to hyperspectral images and can interfere with the feature learning process in the early stage. For band selection, we carefully design an additional parallel input to the attention module for obtaining xed selected band sets and an activation function for ltering insigni cant bands in the training process.
In the third work, we propose attention mechanisms to address multi-domain hyperspectral image classi cation. Di erent hyperspectral datasets have di erent data modalities, statistical distributions, or spectral dimensionalities. This brings signi cant challenges for a single network to learn all the tasks. The domain shift problem can be alleviated by adjusting the network towards the property of speci c domains. To this end, domain attention modules are designed to attend to the domain of the input data for adapting the network accordingly. Two domain attention modules: hard domain attention and soft domain attention are proposed. For the hard domain attention network, the attention mechanism is implemented by a muxer switch. According to the labels of data domain, a set of small domain speci c adapters are selected and connected to a main backbone network. In this way, the majority of network parameters are shared by all domains with only a small number of domain speci c parameters. For the soft domain attention network, we build the attention mechanism based on squeeze and excitation (SE) block. Several parallel SE blocks are applied as the feature adapters. On top of them, a higher level domain attention SE block is placed to achieve domain assignment.Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Info & Comm Tech
Science, Environment, Engineering and Technology
Full Text
6
DAL, MOLIN Anna. « Interaction between mechanism of attention selection in space and time : Behavioural and electrophysiological evidence ». Doctoral thesis, Università degli Studi di Verona, 2009. http://hdl.handle.net/11562/337444.
Texte intégralRésumé :
I meccanismi attentivi consentono di selezionare dall'ambiente circostante le informazioni utili allosvolgimento di un determinato compito. Negli ultimi trenta anni, i processi coinvolti nella selezionedi informazioni di natura spaziale sono stati ampiamente investigati, mentre rimangono ancora dachiarire i meccanismi coinvolti negli aspetti di selezione temporale. I tre esperimenti riportatiall'interno di questa tesi sono volti ad indagare alcuni degli aspetti legati alla capacità di selezionaregli eventi nel tempo ed in che modo gli aspetti temporali e quelli spaziali interagiscono tra loro.Nel primo esperimento è stato impiegato un compito di Giudizio di Ordine Temporale (TOJ) perinvestigare la relazione esistente tra disturbi di selezione nello spazio e nel tempo in pazienti coneminegligenza spaziale unilaterale. Una forte compromissione dei meccanismi di selezione neltempo è stata rilevata per le coppie di stimoli presentate in porzioni dello spazio in cui il deficitspaziale è più marcato, suggerendo l'esistenza una relazione tra gli aspetti spaziali e quelli temporalinella modulazione del deficit.Nel secondo e nel terzo esperimento è stato investigato l'orientamento dell'attenzione nel tempoutilizzando stimoli che, grazie ad un movimento con velocità regolare o irregolare, rendonopossibile il generarsi di aspettative temporali e di verificare cosa avviene quando tali aspettativevengono disattese. La regolarità del movimento si è rivelato essere un indice importante nelgenerare aspettative temporali che a loro volta influenzano profondamente la performancediminuendo sensibilmente la velocità di risposta del soggetto. Inoltre, la registrazione dei potenzialievocati ha evidenziato come aspettative spaziali e temporali interagiscano influenzando l'analisidello stimolo fin dalle prime fasi di elaborazione.The study of mechanisms involved in spatial attention is one of the most investigated field inmodern neuroscience, but in the last years a growing interest has been devoted to unveil themechanisms concerning also the temporal aspects of attention. In this thesis three experiment arereported that tried to cast more light on the temporal aspects of attention and on the relationshipbetween spatial and temporal attentional mechanisms.In the first experiment the relationship between spatial and temporal deficit in selective visualattention has been investigated in a group of neglect patients using a temporal order judgement task(TOJ). The main finding is a stronger impairment in temporal selection for spatial position in whichthe attention selection is more impaired, suggesting an interaction between the two aspects in themodulation of the deficit.The second and the third experiment investigated temporal expectations generated by a regularrhythm. In particular, the impact of exogenous and endogenous temporal expectation has beencompared in a discrimination task, revealing the pervasive effect of regularity of movement andspeed in orienting attention in time. Moreover, it has been confirmed the combined effect of spatialand temporal expectations in modulation of electrophysiological response.These results suggest the existence of an interaction between spatial and temporal mechanisms ofattention.
7
Isunza, Navarro Abgeiba Yaroslava. « Evaluation of Attention Mechanisms for Just-In-Time Software Defect Prediction ». Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-288724.
Texte intégralRésumé :
Just-In-Time Software Defect Prediction (JIT-DP) focuses on predicting errors in software at change-level with the objective of helping developers identify defects while the development process is still ongoing, and improving the quality of software applications. This work studies deep learning techniques by applying attention mechanisms that have been successful in, among others, Natural Language Processing (NLP) tasks. We introduce two networks named Convolutional Neural Network with Bidirectional Attention (BACNN) and Bidirectional Attention Code Network (BACoN) that employ a bi-directional attention mechanism between the code and message of a software change. Furthermore, we examine BERT [17] and RoBERTa [57] attention architectures for JIT-DP. More specifically, we study the effectiveness of the aforementioned attention-based models to predict defective commits compared to the current state of the art, DeepJIT [37] and TLEL [101]. Our experiments evaluate the models by using software changes from the OpenStack open source project. The results showed that attention-based networks outperformed the baseline models in terms of accuracy in the different evaluation settings. The attention-based models, particularly BERT and RoBERTa architectures, demonstrated promising results in identifying defective software changes and proved to be effective in predicting defects in changes of new software releases.Just-In-Time Defect Prediction (JIT-DP) fokuserar på att förutspå fel i mjukvara vid ändringar i koden, med målet att hjälpa utvecklare att identifiera defekter medan utvecklingsprocessen fortfarande är pågående, och att förbättra kvaliteten hos applikationsprogramvara. Detta arbete studerar djupinlärningstekniker genom att tillämpa attentionmekanismer som har varit framgångsrika inom, bland annat, språkteknologi (NLP). Vi introducerar två nätverk vid namn Convolutional Neural Network with Bidirectional Attention (BACNN), och Bidirectional Attention Code Network (BACoN), som använder en tvåriktad attentionmekanism mellan koden och meddelandet om en mjukvaruändring. Dessutom undersöker vi BERT [17] och RoBERTa [57], attentionarkitekturer för JIT-DP. Mer specifikt studerar vi hur effektivt dessa attentionbaserade modeller kan förutspå defekta ändringar, och jämför dem med de bästa tillgängliga arkitekturerna DeePJIT [37] och TLEL [101]. Våra experiment utvärderar modellerna genom att använda mjukvaruändringar från det öppna källkodsprojektet OpenStack. Våra resultat visar att attentionbaserade nätverk överträffar referensmodellen sett till träffsäkerheten i de olika scenarierna. De attentionbaserade modellerna, framför allt BERT och RoBERTa, demonstrerade lovade resultat när det kommer till att identifiera defekta mjukvaruändringar och visade sig vara effektiva på att förutspå defekter i ändringar av nya mjukvaruversioner.
8
PUTELLI, LUCA. « Attention Mechanism e Interpretabilità del Deep Learning per il Natural Language Processing in Ambito Biomedico ». Doctoral thesis, Università degli studi di Brescia, 2021. http://hdl.handle.net/11379/548259.
Texte intégral
9
Raykos, Bronwyn C. « Attentional and interpretive biases : independent dimensions of individual difference or expressions of a common selective processing mechanism ? / ». Connect to this title, 2006. http://theses.library.uwa.edu.au/adt-WU2007.0018.
Texte intégral
10
Ma, Tengfei. « A Graph Attention plus Reinforcement Learning Method for Antenna Tilt Optimization ». Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-300111.
Texte intégralRésumé :
Remote Electrical Tilt optimization is an effective method to obtain the optimal Key Performance Indicators (KPIs) by remotely controlling the base station antenna’s vertical tilt. To improve the KPIs aims to improve antennas’ cooperation effect since KPIs measure the quality of cooperation between the antenna to be optimized and its neighbor antennas. Reinforcement Learning (RL) is an appropriate method to learn an antenna tilt control policy since the agent in RL can generate the optimal epsilon greedy tilt optimization policy by observing the environment and learning from the state- action pairs. However, existing models only produced tilt modification strategies by interpreting the to- be- optimized antenna’s features, which cannot fully characterize the mobile cellular network formed by the to- be- optimized antenna and its neighbors. Therefore, incorporating the features of the neighboring antennas into the model is an important measure to improve the optimization strategy. This work will introduce the Graph Attention Network to model the neighborhood antenna’s impact on the antenna to be optimized through the attention mechanism. Furthermore, it will generate a low- dimensional embedding vector with more expressive power to represent the to- be- optimized antenna’s state in the RL framework through dealing with graph- structural data. This new model, namely Graph Attention Q- Network (GAQ), is a model based on DQN and aims to acquire a higher performance than the Deep Q- Network (DQN) model, which is the baseline, evaluated by the same metric — KPI Improvement. Since GAQ has a richer perception of the environment than the vanilla DQN model, it thereby outperforms the DQN model, obtaining fourteen percent performance improvement compared to the baseline. Besides, GAQ also performs 14 per cent better than DQN in terms of convergence efficiency.Optimering av fjärrlutning är en effektiv metod för att nå optimala nyckeltal genom fjärrstyrning av den vertikala lutningen av en antenn i en basstation. Att förbättra nyckeltalen innebär att förbättra sammarbetseffekten mellan antenner eftersom nyckeltalen är mått på kvalitén av sammarbetet mellan den antenn som optimeras och dess angränsande antenner. Förstärkande Inlärning (FI) är en lämplig metod för att lära sig en optimal strategi för reglering av antennlutningen eftersom agenten inom FI kan generera den optimala epsilongiriga optimeringsstrategin genom att observera miljön och lära sig från par av tillstånd och aktioner. Nuvarande modeller genererar dock endast lutningsstrategier genom att tolka egenskaperna hos den antenn som ska optimeras, vilket inte är tillräckligt för att karatärisera mobilnätverket bestående av antennen som ska optimeras samt dess angränsande antenner. Därav är inkluderingen av de angränsande antennernas egenskaper i modellen viktig för att förbättra optimeringsstrategin. Detta arbete introducerar Graf- Uppmärksammat Nätverk för att modellera de angränsande antennernas påverkan på den antenn som ska optimeras genom uppmärksamhetsmekanismen. Metoden genererar en lågdimensionell vektor med större förmåga att representera den optimerade antennens tillstånd i FI modellen genom att hantera data i struktur av en graf. Den nya modellen, Graf- Uppmärksammat Q- Nätverk (GUQ), är en modell baserad på DQN med mål att nå bättre prestanda än en standard DQN- modell, utvärderat efter samma mätvärde –– förbättring av nyckeltalen. Eftersom GUQ har en större upfattning av miljön så överträffar metoden DQN- modellen genom en fjorton procent bättre prestandaökning. Dessutom, så överträffar GUQ även DQN i form av snabbare konvergens.
11
Dou, Tianyu. « Multi-Kernel Deformable 3D Convolution for Video Super-Resolution ». Thesis, Université d'Ottawa / University of Ottawa, 2021. http://hdl.handle.net/10393/42682.
Texte intégralRésumé :
Video super-resolution (VSR) methods align and fuse consecutive low-resolution frames to generate high-resolution frames. One of the main difficulties for the VSR process is that video contains various motions, and the accuracy of motion estimation dramatically affects the quality of video restoration. However, standard CNNs share the same receptive field in each layer, and it is challenging to estimate diverse motions effectively. Neuroscience research has shown that the receptive fields of biological visual areas will be adjusted according to the input information. Diverse receptive fields in temporal and spatial dimensions have the potential to adapt to various motions, which is rarely paid attention in most known VSR methods.
In this thesis, we propose to provide adaptive receptive fields for the VSR model. Firstly, we design a multi-kernel 3D convolution network and integrate it with a multi-kernel deformable convolution network for motion estimation and multiple frames alignment. Secondly, we propose a 2D multi-kernel convolution framework to improve texture restoration quality. Our experimental results show that the proposed framework outperforms the state-of-the-art VSR methods.
12
Chen, Meihong. « Real-Time Video Object Detection with Temporal Feature Aggregation ». Thesis, Université d'Ottawa / University of Ottawa, 2021. http://hdl.handle.net/10393/42790.
Texte intégralRésumé :
In recent years, various high-performance networks have been proposed for single-image object detection. An obvious choice is to design a video detection network based on state-of-the-art single-image detectors. However, video object detection is still challenging due to the lower quality of individual frames in a video, and hence the need to include temporal information for high-quality detection results. In this thesis, we design a novel interleaved architecture combining a 2D convolutional network and a 3D temporal network. We utilize Yolov3 as the base detector. To explore inter-frame information, we propose feature aggregation based on a temporal network. Our temporal network utilizes Appearance-preserving 3D convolution (AP3D) for extracting aligned features in the temporal dimension. Our multi-scale detector and multi-scale temporal network communicate at each scale and also across scales. The number of inputs of our temporal network can be either 4, 8, or 16 frames in this thesis and correspondingly we name our temporal network TemporalNet-4, TemporalNet-8 and TemporalNet-16. Our approach achieves 77.1\% mAP (mean Average Precision) on ImageNet VID 2017 dataset with TemporalNet-4, where TemporalNet-16 achieves 80.9\% mAP which is a competitive result on this video object detection benchmark. Our network is also real-time with a running time of 35ms/frame.
13
Dronzeková, Michaela. « Analýza polygonálních modelů pomocí neuronových sítí ». Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2020. http://www.nusl.cz/ntk/nusl-417253.
Texte intégralRésumé :
This thesis deals with rotation estimation of 3D model of human jaw. It describes and compares methods for direct analysis od 3D models as well as method to analyze model using rasterization. To evaluate perfomance of proposed method, a metric that computes number of cases when prediction was less than 30° from ground truth is used. Proposed method that uses rasterization, takes three x-ray views of model as an input and processes it with convolutional network. It achieves best preformance, 99% with described metric. Method to directly analyze polygonal model as a sequence uses attention mechanism to do so and was inspired by transformer architecture. A special pooling function was proposed for this network that decreases memory requirements of the network. This method achieves 88%, but does not use rasterization and can process polygonal model directly. It is not as good as rasterization method with x-ray display, byt it is better than rasterization method with model not rendered as x-ray. The last method uses graph representation of mesh. Graph network had problems with overfitting, that is why it did not get good results and I think this method is not very suitable for analyzing plygonal model.
14
Shi, Fangzhou. « Towards Molecule Generation with Heterogeneous States via Reinforcement Learning ». Thesis, The University of Sydney, 2020. https://hdl.handle.net/2123/22335.
Texte intégralRésumé :
De novo molecular design and generation are frequently prescribed in the field of chemistry and biology, for it plays a critical role in maintaining the prosperity of the chemical industry and benefiting the drug discovery. Nowadays, many significant problems in this field are based on the philosophy of designing molecular structures towards specific desired properties. This research is very meaningful in both medical and AI fields, which can benefits novel drug discovery for some diseases. However, It remains a challenging task due to the large size of chemical space. In recent years, reinforcement learning-based methods leverage graphs to represent molecules and generate molecules as a decision making process. However, this vanilla graph representation may neglect the intrinsic context information with molecules and limits the generation performance accordingly. In this paper, we propose to augment the original graph states with the SMILES context vectors. As a result, SMILES representations are easily processed by a simple language model such that the general semantic features of a molecule can be extracted; and the graph representations perform better in handling the topology relationship of each atom. Moreover, we propose a framework that combines supervised learning and reinforcement learning algorithm to take a solid consideration of these two heterogeneous state representations of a molecule, which can fuse the information from both of them and extract more comprehensive features so that more sophisticated decisions can be made by the policy network. Our model also introduces two attention mechanisms, i.e., action-attention, and graph-attention, to further improve the performance. We conduct our experiments on a practical dataset, ZINC, and the experiment results demonstrate that our framework can outperform other baselines in the learning performance of molecule generation and chemical property optimization.
15
Ahmad, Fawad. « Role of Cognitive Processes, Emotional Regulation, Attention, and Intrinsic Motives in explaining the underlying Mechanism and Dynamics of Value Premium : A Mispricing Perspective ». Doctoral thesis, Luiss Guido Carli, 2020. http://hdl.handle.net/11385/203452.
Texte intégralRésumé :
Is the investor’s reliance on cognitive processes and emotional regulation strategy predict preferences towards the selection of value versus growth stocks? Is the value premium vary across the level of investors’ attention? Is the value premium dependent on mispricing signals manifested in the firms’ intangibles-intensity? Investor’s Intrinsic Motives and the Valence of Word-of-Mouth in Sequential Decision-Making.
16
Lee, Yun K. « Unveiling the underlying mechanism for the matching effect between construal level and message frames : how and why do matches between gain versus loss frames and construal level enhance persuasion ? » Diss., University of Iowa, 2012. https://ir.uiowa.edu/etd/3332.
Texte intégralRésumé :
The current research investigates how and why consumers' construal levels and the appeals framed either by gains or losses jointly influence persuasion. The findings across four experiments indicate that matching high-level construals with gain frames and low-level construals with loss frames leads to a) higher intentions to engage in cholesterol lowering behavior (experiment 1), b) more favorable brand attitudes (experiment 2), c) greater willingness to donate to an environmental organization (experiment 3), and d) higher buying intentions for a brand (experiment 4). It seems that these outcomes occur because matches between construal level and message frames encourage people to pay attention to the information they evaluate (experiments 1 ˜4), and this enhanced attention induces greater perceptions of processing fluency, which in turn leads to positive attitudes (experiments 2˜4). Further, this research demonstrates that an adequate amount of cognitive resources is required for this matching effect to occur (experiment 4). The current research contributes to the construal level, message framing, and matching literatures by unveiling the specific mechanism underlying the matching relationship between construal level and gain versus loss frames on persuasion and by identifying a boundary condition for it. This research also has managerial implications for marketing managers and policymakers in that it suggests a strategic way to use construal level and message frames to enhance marketing communication and advertising effectiveness.
17
Wilson, Steven. « Mechanisms of attention and awareness : parameters and assessment of pre-attentional awareness using change blindness task ». Thesis, University of Lincoln, 2008. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.496078.
Texte intégral
18
Kerr, John H. « Arousal mechanisms, attention and sports performance ». Thesis, University of Nottingham, 1988. http://eprints.nottingham.ac.uk/10947/.
Texte intégralRésumé :
This thesis is concerned with the relationship between arousal mechanisms, attentional processes and competitive sports performance. Theoretical interpretations of the arousal-performance relationship have traditionally followed the inverted-U hypothesis. Based on this approach, the generally accepted view in sports psychology is that high levels of arousal are detrimental to good performance. A review of the relevant psychological literature reveals the limited nature of such an approach and draws attention to alternative perspectives such as those offered by the work of Apter and that of Cox and Mackay. These more recent theoretical approaches allow more sophisticated interpretations of the individual's experience of arousal to be realised. Important here are other aspects of the individual's psychological state (cognition and emotion) as these are thought to affect his or her interpretation of arousal. Interestingly, the two theories, developed independently by Apter and by Cox and Mackay, appear consistent, one with the other, and have not previously been applied to the study of competitive sport. Several different research techniques were incorporated into a research design which used squash players of varying levels of ability to examine the various psychological factors important in their experience of and performance in competitive squash. The research techniques, some of which were innovative, proved effective indentifying the interaction of arousal and stress in relation to competitive performance. It was concluded that psychological preparation and experience (i.e. number of years, number of times per week played), along with personality characteristics and attentional strategies, contribute to success in competitive squash. Fluctuations in emotional responses characterised players whose performance was unsuccessful. By way of contrast, successful players' (i.e. successful in terms of level of ability attained, skill performance and winning games) psychological responses were more consistent. They achieved and maintained high levels of arousal both prior to and during performance. High arousal was, for successful players, accompanied by low stress and positive hedonic tone when they were subject to the demands of competitive squash games. Overall, successful players (that is skilled players in Study 2 and winners from Study 3) were highly extravert and significantly less neurotic (Eysenck) than other groups of players. Telic dominance was not a discriminating characteristic in this investigation, but successful players' attentional styles were significantly different, as defined by Nideffer's BIT and INFP subscales, to those styles or strategies employed by less capable players. Successful players generally employed psychological preparation strategies prior to and during play to a greater extent than other players. When doing so, they were more concerned with cognitive strategies, in the form of focussing and planning, than arousal modulation strategies. The present research investigation advanced knowledge about the processes involved in competitive sports, providing new and relevant information. As a result, a number of suggestions for squash coaching and player development, along with implications for cognitive intervention with sports performers, have emerged.
19
Gama, Nuno. « Mechanisms of multisensory integration and attention ». Thesis, University of Nottingham, 2017. http://eprints.nottingham.ac.uk/40400/.
Texte intégralRésumé :
Spatial attention is an essential mechanism that helps us perceive our surroundings by bringing into consciousness environmental occurrences or objects that may be of importance. Studies of spatial attention have classically recorded behavioural responses to targets presented in a region of space where attention had previously been allocated to. Such investigations show a behavioural facilitation at the same location due to cueing, but less in known about the effects of shifts of attention when the cued location is not the location of interest. This thesis presents seven experiments aimed at investigating this by implementing and revising the attentional repulsion effect (ARE). The ARE is a perceptual localisation error when attention is diverted from the region of interest and it has been extensively studied in the visual domain, however, the rising number of ARE studies has created numerous research methodologies used to evoke the effect, which may have led to isolated reports. This thesis attempts to combine past methodologies with a new approach to quantify the effect, and will address some methodological differences evident in the literature, in order to optimise the stimulus paradigms and maximise the effect. The results show that a robust ARE can be elicited in the visual modality, but the same is not observed in the auditory modality. Furthermore, when using cues that are of different modality than the targets, the ARE is only observed in the visual target modality. Using visual cues and auditory targets will produce an attraction effect, in line with the ventriloquism theory. However, the implementation of interstimuli intervals up to 1.5 seconds would be enough to disrupt the ventriloquism illusion, but it did not alter the resulted attraction. Lastly, one question regarding the role of attention in sensory adaptation was addressed. I hypothesise that sensory adaptation could be further a contributor to the ARE given that most psychophysics paradigms of the ARE repeat the same stimuli thousands of times, uninterruptedly. The results are inconclusive mainly due to experimental design. All results are discussed in relation with theories of spatial and multimodal attention.
20
Furlan, Michele. « Neural mechanisms of attention to motion ». Doctoral thesis, Università degli studi di Trento, 2011. https://hdl.handle.net/11572/368674.
Texte intégralRésumé :
Attention is thought to alter appearance by intensifying the sensory impression of the attended stimulus. Current debates are about how this increases might operate and what are the underlying neural mechanisms. Three mechanisms have been proposed to account the effect of attention: contrast gain, response gain and baseline shift. However, psychophysical and neuroimaging studies produced results that are not always consistent. We used the functional magnetic resonance imaging (fMRI) to explore how attention alters the appearance of coherently moving dots. The first experiment assessed coherence response functions for different levels of contrast. This experiment showed which kind of pattern should be expected with attention altering appearance by means of contrast gain mechanism. The results showed that contrast produced a leftward shift of the coherence response function. By our knowledge, this is the first fMRI evidence for the assumption that processing of motion coherence is susceptible to contrast gain. The second experiment assessed the effect of attention on coherence response function. The results showed that attention affects the BOLD response with the same strength for all coherence levels. This pattern of response is compatible with the predictions of the baseline shift model. Finally, in order to explain the discrepancy between psychophysical and neuroimaging results, we suggested a signal detection account for the effect of baseline shift on the perceptual level, proposing that baseline shift may produce a leftward shift of the psychometric function although sensitivity remains unchanged.
21
Galashan, Fingal Orlando [Verfasser], Detlef [Akademischer Betreuer] Wegener, Andreas [Akademischer Betreuer] Kreiter et Michael [Akademischer Betreuer] Koch. « Selective visual attention : A mechanism for optimal adjustment of sensory processing to task requirements - from method development to human psychophysics and monkey single cell recordings / Fingal Orlando Galashan. Gutachter : Andreas Kreiter ; Michael Koch. Betreuer : Detlef Wegener ». Bremen : Staats- und Universitätsbibliothek Bremen, 2011. http://d-nb.info/1071842129/34.
Texte intégral
22
Owen, Adrian Mark. « Fronto-striatal mechanisms in planning and attention ». Thesis, King's College London (University of London), 1992. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.302625.
Texte intégral
23
Gontrum, Johannes. « Attention Mechanisms for Transition-based Dependency Parsing ». Thesis, Uppsala universitet, Institutionen för lingvistik och filologi, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-395491.
Texte intégralRésumé :
Transition-based dependency parsing is known to compute the syntactic structure of a sentence efficiently, but is less accurate to predict long-distance relations between tokens as it lacks global information about the sentence. Our main contribution is the integration of attention mechanisms to replace the static token selection with a dynamic approach that takes the complete sequence into account. Though our experiments confirm that our approach fundamentally works, our models do not outperform the baseline parser. We further present a line of follow-up experiments to investigate these results. Our main conclusion is that the BiLSTM of the traditional parser is already powerful enough to encode the required global information into each token, eliminating the need for an attention-driven approach. Our secondary results indicate that the attention models require a neural network with a higher capacity to potentially extract more latent information from the word embeddings and the LSTM than the traditional parser. We further show that positional encodings are not useful for our attention models, though BERT-style positional embeddings slightly improve the results. Finally, we experiment with replacing the LSTM with a Transformer-encoder to test the impact of self-attention. The results are disappointing, though we think that more future research should be dedicated to this. For our work, we implement a UUParser-inspired dependency parser from scratch in PyTorch and extend it with, among other things, full GPU support and mini-batch processing. We publish the code under a permissive open source license at https://github.com/jgontrum/parseridge.
24
Santangelo, Valerio. « Multimodal investigation on spatial attention mechanisms : a model of shared attention resources (ShAR) ». Doctoral thesis, La Sapienza, 2005. http://hdl.handle.net/11573/917227.
Texte intégral
25
Kennett, Steffan Anthony. « Links in spatial attention between touch and vision ». Thesis, Birkbeck (University of London), 2000. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.343623.
Texte intégral
26
Das, Srijan. « Mécanismes d'attention spatio-temporels pour la reconnaissance d'activité ». Thesis, Université Côte d'Azur, 2020. https://tel.archives-ouvertes.fr/tel-03177892.
Texte intégralRésumé :
Cette thèse vise la reconnaissance des actions humaines dans les vidéos. La reconnaissance des actions est une tâche compliquée dans le domaine de la vision par ordinateur en raison de ses défis complexes et élevés. Avec l'émergence de l'apprentissage en profondeur et des ensembles de données à grande échelle provenant de sources Internet, des améliorations substantielles ont été apportées à la compréhension vidéo. Par exemple, des réseaux convolutifs 3D de pointe comme I3D pré-formés sur d'énormes ensembles de données comme Kinetics ont réussi à accroître la reconnaissance des actions des vidéos Internet. Mais, ces réseaux à noyaux rigides appliqués sur l'ensemble du volume espace-temps ne peuvent pas relever les défis présentés par les activités de la vie quotidienne (ADL).Nous sommes particulièrement intéressés par la représentation vidéo discriminante pour ADL. Outre les défis des vidéos génériques, ADL présente - (i) des actions à grain fin avec des mouvements courts et subtils comme verser du grain et verser de l'eau, (ii) des actions avec des modèles visuels similaires différant par des modèles de mouvement comme se frotter les mains et applaudir, et enfin ( iii) de longues actions complexes comme la cuisine. Afin de relever ces défis, nous avons apporté des contributions.La première contribution comprend - une stratégie de fusion multimodale pour prendre en compte les avantages des modalités multiples pour classer les actions. Cependant, la question demeure: comment combiner plusieurs modalités de bout en bout? Comment pouvons-nous utiliser les informations 3D pour guider les réseaux RVB de pointe actuels pour la classification des actions? À cette fin, nous proposons des mécanismes articulés d'attention axée sur la pose pour la classification des actions. Nous proposons trois variantes de mécanismes d'attention spatio-temporelle exploitant les modalités de pose RVB et 3D pour relever les défis susmentionnés (i) et (ii) pour des actions courtes. Notre troisième contribution principale est un modèle temporel en plus de notre modèle basé sur l'attention. La représentation vidéo conservant des informations temporelles denses permet au modèle temporel de modéliser de longues actions complexes, ce qui est crucial pour ADL.Nous avons évalué notre première contribution sur trois ensembles de données publics à petite échelle: CAD-60, CAD-120 et MSRDailyActivity3D. D'un autre côté, nous avons évalué nos deux contributions restantes sur quatre ensembles de données publics: un ensemble de données sur l'activité humaine à grande échelle: NTU-RGB + D 120, son sous-ensemble NTU-RGB + D 60, un ensemble de données sur l'activité humaine difficile dans le monde réel: Toyota Smarthome et un ensemble de données d'interaction homme-objet à petite échelle Northwestern UCLA. Nos expériences montrent que les méthodes proposées dans cette thèse surpassent les résultats de pointeThis thesis targets recognition of human actions in videos. Action recognition is a complicated task in the field of computer vision due to its high complex challenges. With the emergence of deep learning and large scale datasets from internet sources, substantial improvements have been made in video understanding. For instance, state-of-the-art 3D convolutional networks like I3D pre-trained on huge datasets like Kinetics have successfully boosted the recognition of actions from internet videos. But, these networks with rigid kernels applied across the whole space-time volume cannot address the challenges exhibited by Activities of Daily Living (ADL). We are particularly interested in discriminative video representation for ADL. Besides the challenges in generic videos, ADL exhibits - (i) fine-grained actions with short and subtle motion like pouring grain and pouring water, (ii) actions with similar visual patterns differing in motion patterns like rubbing hands and clapping, and finally (iii) long complex actions like cooking. In order to address these challenges, we have made three key contributions. The first contribution includes - a multi-modal fusion strategy to take the benefits of multiple modalities into account for classifying actions. However the question remains, how to combine multiple modalities in an end-to-end manner? How can we make use of the 3D information to guide the current state-of-the-art RGB networks for action classification? To this end, we propose articulated pose driven attention mechanisms for action classification. We propose, three variants of spatio-temporal attention mechanisms exploiting RGB and 3D pose modalities to address the aforementioned challenges (i) and (ii) for short actions. Our third main contribution is a Temporal Model on top of our attention based model. The video representation retaining dense temporal information enables the temporal model to model long complex actions which is crucial for ADL.We have evaluated our first contribution on three small-scale public datasets: CAD-60, CAD-120 and MSRDailyActivity3D. On the other hand, we have evaluated our remaining two contributions on four public datasets: a large scale human activity dataset: NTU-RGB+D 120, its subset NTU-RGB+D 60, a real-world challenging human activity dataset: Toyota Smarthome and a small scale human-object interaction dataset Northwestern UCLA. Our experiments show that the methods proposed in this thesis outperform the state-of-the-art results
27
Tattersall, A. J. « Divided attention and the structure of temporary memory ». Thesis, University of Oxford, 1987. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.382717.
Texte intégral
28
Battistoni, Elisa. « Attentional Mechanisms in Natural Scenes ». Doctoral thesis, Università degli studi di Trento, 2018. https://hdl.handle.net/11572/369086.
Texte intégralRésumé :
The visual analysis of the world around us is an incredibly complex neural process that allows humans to function appropriately within the environment. When one considers the intricacy of both the visual input and the (currently known) neural mechanisms necessary for its analysis, it is difficult not to remain enchanted by the fact that, even though the signal that hits the retina has a tremendous amount of simple visual features and that is ever changing, ambiguous and incomplete, we experience the world around us in a very easy, stable and straightforward manner. So much effort has been put into the study of vision, and despite the enormous scientific advances and important findings, many questions still need answers.
During my years spent as Ph.D. student, I investigated some questions related to top-down attentional mechanisms in real-world visual search. Specifically, Chapter 2 and 3 address the processing stage of preparation, by investigating the characteristics of attentional templates when preparing to search for objects in scenes; Chapter 4 addresses the stage of guidance and selection, by investigating the temporal course of spatial attention guidance; and finally, Chapter 5 addresses the identification phase, by investigating the temporal dynamics of size-constancy mechanisms in real-world scenes.
To anticipate some results, we proposed that attentional templates in real-world visual search tasks are based on category-diagnostic features and code the expected target size/distance. In the context of the attentional guidance and selection stage, we demonstrate that attention spatially focuses on targets around 240ms, following category-based attentional modulations appearing at 180ms after scene onset. Finally, we propose that size constancy mechanisms appear before 200ms post-scene. This is in line with the expectation that a coarse identification of an object, including its size, should be computed before spatially focusing attention onto the target. Together these studies improve our understanding of top-down attentional processes engaged in real-world visual search, and raise some questions which future research could address.
29
Paltoglou, Aspasia Eleni. « Mechanisms of spatial and non-spatial auditory selective attention ». Thesis, University of Nottingham, 2009. http://eprints.nottingham.ac.uk/10697/.
Texte intégralRésumé :
Selective attention is a crucial function that encompasses all perceptual modalities and which enables us to focus on the behaviorally relevant information and ignore the rest. The main goal of the thesis is to test well-established hypotheses about the mechanisms of visual selective attention in the auditory domain using behavioral and neuroimaging methods. Two fMRI studies (Experiments 1 and 2) test the hypothesis of feature-specific attentional enhancement. This hypothesis states that when attending to an object or a feature, there should be an enhancement of the response in the sensory region that is sensitive to that object or feature. Experiment 1 investigated feature-specific attentional modulation mainly within the tonotopic fields around primary auditory cortex. Experiment 2 investigated feature-specific attentional modulation mainly around non-primary auditory cortex, when attending to frequency modulation or motion of the same auditory object. Experiment 1 showed evidence for feature-specific enhancement, while Experiment 2 did not. The role of competition among concurrent auditory objects as a necessary factor in driving feature-specific enhancement is discussed. A second hypothesis from vision research is that spatial perception and attention is much more precise in the centre than in the periphery. Experiment 3 used a masking release paradigm to investigate whether the acuity of auditory spatial attention was similarly increased in the midline. Although location discrimination of sounds segregated by inter-aural time differences was more precise at the midline than at the periphery, spatial attention was not. Therefore for this task at least there was no effect of eccentricity on auditory spatial attention. The results of these three studies are discussed in view of selective attention as a flexible process that operates in different ways according to the specifics of the task.
30
Beuth, Frederik. « Visual attention in primates and for machines - neuronal mechanisms ». Universitätsverlag Chemnitz, 2017. https://monarch.qucosa.de/id/qucosa%3A35655.
Texte intégralRésumé :
Visual attention is an important cognitive concept for the daily life of humans, but still not fully understood. Due to this, it is also rarely utilized in computer vision systems. However, understanding visual attention is challenging as it has many and seemingly-different aspects, both at neuronal and behavioral level. Thus, it is very hard to give a uniform explanation of visual attention that can account for all aspects. To tackle this problem, this thesis has the goal to identify a common set of neuronal mechanisms, which underlie both neuronal and behavioral aspects. The mechanisms are simulated by neuro-computational models, thus, resulting in a single modeling approach to explain a wide range of phenomena at once. In the thesis, the chosen aspects are multiple neurophysiological effects, real-world object localization, and a visual masking paradigm (OSM). In each of the considered fields, the work also advances the current state-of-the-art to better understand this aspect of attention itself. The three chosen aspects highlight that the approach can account for crucial neurophysiological, functional, and behavioral properties, thus the mechanisms might constitute the general neuronal substrate of visual attention in the cortex. As outlook, our work provides for computer vision a deeper understanding and a concrete prototype of attention to incorporate this crucial aspect of human perception in future systems.:1. General introduction
2. The state-of-the-art in modeling visual attention
3. Microcircuit model of attention
4. Object localization with a model of visual attention
5. Object substitution masking
6. General conclusionVisuelle Aufmerksamkeit ist ein wichtiges kognitives Konzept für das tägliche Leben des Menschen. Es ist aber immer noch nicht komplett verstanden, so dass es ein langjähriges Ziel der Neurowissenschaften ist, das Phänomen grundlegend zu durchdringen. Gleichzeitig wird es aufgrund des mangelnden Verständnisses nur selten in maschinellen Sehsystemen in der Informatik eingesetzt. Das Verständnis von visueller Aufmerksamkeit ist jedoch eine komplexe Herausforderung, da Aufmerksamkeit äußerst vielfältige und scheinbar unterschiedliche Aspekte besitzt. Sie verändert multipel sowohl die neuronalen Feuerraten als auch das menschliche Verhalten. Daher ist es sehr schwierig, eine einheitliche Erklärung von visueller Aufmerksamkeit zu finden, welche für alle Aspekte gleichermaßen gilt. Um dieses Problem anzugehen, hat diese Arbeit das Ziel, einen gemeinsamen Satz neuronaler Mechanismen zu identifizieren, welche sowohl den neuronalen als auch den verhaltenstechnischen Aspekten zugrunde liegen. Die Mechanismen werden in neuro-computationalen Modellen simuliert, wodurch ein einzelnes Modellierungsframework entsteht, welches zum ersten Mal viele und verschiedenste Phänomene von visueller Aufmerksamkeit auf einmal erklären kann. Als Aspekte wurden in dieser Dissertation multiple neurophysiologische Effekte, Realwelt Objektlokalisation und ein visuelles Maskierungsparadigma (OSM) gewählt. In jedem dieser betrachteten Felder wird gleichzeitig der State-of-the-Art verbessert, um auch diesen Teilbereich von Aufmerksamkeit selbst besser zu verstehen. Die drei gewählten Gebiete zeigen, dass der Ansatz grundlegende neurophysiologische, funktionale und verhaltensbezogene Eigenschaften von visueller Aufmerksamkeit erklären kann. Da die gefundenen Mechanismen somit ausreichend sind, das Phänomen so umfassend zu erklären, könnten die Mechanismen vielleicht sogar das essentielle neuronale Substrat von visueller Aufmerksamkeit im Cortex darstellen. Für die Informatik stellt die Arbeit damit ein tiefergehendes Verständnis von visueller Aufmerksamkeit dar. Darüber hinaus liefert das Framework mit seinen neuronalen Mechanismen sogar eine Referenzimplementierung um Aufmerksamkeit in zukünftige Systeme integrieren zu können. Aufmerksamkeit könnte laut der vorliegenden Forschung sehr nützlich für diese sein, da es im Gehirn eine Aufgabenspezifische Optimierung des visuellen Systems bereitstellt. Dieser Aspekt menschlicher Wahrnehmung fehlt meist in den aktuellen, starken Computervisionssystemen, so dass eine Integration in aktuelle Systeme deren Leistung sprunghaft erhöhen und eine neue Klasse definieren dürfte.:1. General introduction 2. The state-of-the-art in modeling visual attention 3. Microcircuit model of attention 4. Object localization with a model of visual attention 5. Object substitution masking 6. General conclusion
31
Greco, Claudio. « Transfer Learning and Attention Mechanisms in a Multimodal Setting ». Doctoral thesis, Università degli studi di Trento, 2022. http://hdl.handle.net/11572/341874.
Texte intégralRésumé :
Humans are able to develop a solid knowledge of the world around them: they can leverage information coming from different sources (e.g., language, vision), focus on the most relevant information from the input they receive in a given life situation, and exploit what they have learned before without forgetting it. In the field of Artificial Intelligence and Computational Linguistics, replicating these human abilities in artificial models is a major challenge. Recently, models based on pre-training and on attention mechanisms, namely pre-trained multimodal Transformers, have been developed. They seem to perform tasks surprisingly well compared to other computational models in multiple contexts. They simulate a human-like cognition in that they supposedly rely on previously acquired knowledge (transfer learning) and focus on the most important information (attention mechanisms) of the input. Nevertheless, we still do not know whether these models can deal with multimodal tasks that require merging different types of information simultaneously to be solved, as humans would do. This thesis attempts to fill this crucial gap in our knowledge of multimodal models by investigating the ability of pre-trained Transformers to encode multimodal information; and the ability of attention-based models to remember how to deal with previously-solved tasks. With regards to pre-trained Transformers, we focused on their ability to rely on pre-training and on attention while dealing with tasks requiring to merge information coming from language and vision. More precisely, we investigate if pre-trained multimodal Transformers are able to understand the internal structure of a dialogue (e.g., organization of the turns); to effectively solve complex spatial questions requiring to process different spatial elements (e.g., regions of the image, proximity between elements, etc.); and to make predictions based on complementary multimodal cues (e.g., guessing the most plausible action by leveraging the content of a sentence and of an image). The results of this thesis indicate that pre-trained Transformers outperform other models. Indeed, they are able to some extent to integrate complementary multimodal information; they manage to pinpoint both the relevant turns in a dialogue and the most important regions in an image. These results suggest that pre-training and attention play a key role in pre-trained Transformers’ encoding. Nevertheless, their way of processing information cannot be considered as human-like. Indeed, when compared to humans, they struggle (as non-pre-trained models do) to understand negative answers, to merge spatial information in difficult questions, and to predict actions based on complementary linguistic and visual cues. With regards to attention-based models, we found out that these kinds of models tend to forget what they have learned in previously-solved tasks. However, training these models on easy tasks before more complex ones seems to mitigate this catastrophic forgetting phenomenon. These results indicate that, at least in this context, attention-based models (and, supposedly, pre-trained Transformers too) are sensitive to tasks’ order. A better control of this variable may therefore help multimodal models learn sequentially and continuously as humans do.
32
Elshafei, Hesham. « Neurophysiological Mechanisms of Auditory Distractibility in the Healthy, Aging or Damaged Human Brain ». Thesis, Lyon, 2018. http://www.theses.fr/2018LYSE1255/document.
Texte intégralRésumé :
Les mécanismes volontaires (V) et involontaires (I) de l’attention reposent sur les réseaux dorsal et ventral, convergeant dans le cortex préfrontal latéral (lPFC). La distractibilité accrue liée au vieillissement ou à une lésion frontale pourrait être due à une altération de l’équilibre entre ces mécanismes V et I, essentiel mais rarement étudié. Notre objectif est de tester, dans la modalité auditive, si (1) les oscillations alpha coordonnent l'activité du réseau dorsal, (2) les oscillations gamma celle du réseau ventral, (3) le couplage oscillatoire dans le lPFC maintient l’équilibre entre les deux réseaux. Ce travail vise également à étudier les corrélats oscillatoires de la distractibilité accrue liée au vieillissement ou à une atteinte frontale. Des données MEEG ont été enregistrées alors que des participants réalisaient le Competitive Attention Test, qui permet d’étudier simultanément les mécanismes V et I de l’attention. Nous avons montré que les oscillations alpha reflètent l’activation des mécanismes facilitateurs et suppresseurs de l’attention V, et la communication au sein du réseau dorsal ; alors que les oscillations gamma indexent l’activation du réseau ventral. De plus, le lPFC serait impliqué dans la communication au sein des deux réseaux, et le PFC médian dans l’équilibre attentionnel V/I. Nous avons également montré que la distractibilité accrue était liée à un déficit d’attention V au cours du vieillissement, et à une altération des processus V et I après lésion frontale. Ce travail de thèse offre donc une meilleure compréhension de la dynamique cérébrale oscillatoire sur laquelle repose l'équilibre attentionnel V/I, et donc la distractibilitéTop-down (TD) and bottom-up (BU) mechanisms of attention are supported by dorsal and ventral networks that mainly overlap in the lateral prefrontal cortex (lPFC). A balance between these mechanisms is essential, yet rarely investigated. Increased distractibility observed during ageing or after frontal damage could result from jeopardizing this balance. It has been proposed that distinct oscillatory frequencies support the activation of these two attention networks. Our main aim was to test, in the auditory modality, whether (1) alpha oscillations would coordinate activity within the dorsal TD network, (2) gamma activity would index the activation of the ventral BU network, (3) the lPFC would support the balance between these networks through oscillatory coupling. We also aimed to investigate the oscillatory correlates of the increased distractibility associated with ageing or frontal damage. MEEG data were recorded while participants performed the Competitive Attention Test, which enables simultaneous investigation of BU and TD attention mechanisms. We showed that alpha oscillations indexed facilitatory and suppressive mechanisms of TD attention, and communication within the dorsal network; while gamma oscillations indexed the ventral network activation. Moreover, the lPFC subtended communication in the two networks; with the TD/BU interaction occurring in the medial PFC. We also showed that ageing-related distractibility was of TD deficit origin. Finally, preliminary results suggest that lPFC damage can impact both TD and BU attention. This thesis provides novel insights into the brain oscillatory dynamics of the TD/BU attentional balance supporting distractibility
33
Magimairaj, Beula M. « Attentional Mechanisms in Children’s Complex Memory Span Performance ». Ohio University / OhioLINK, 2010. http://rave.ohiolink.edu/etdc/view?acc_num=ohiou1267650640.
Texte intégral
34
Hahn, Britta. « Mechanisms of nicotine-induced attentional enhancement ». Thesis, King's College London (University of London), 2002. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.400578.
Texte intégral
35
Cohen, Jason C. « Attention mechanisms and inhibition of return in the somatosensory system ». Related electronic resource : Current Research at SU : database of SU dissertations, recent titles available full text, 2002. http://wwwlib.umi.com/cr/syr/main.
Texte intégral
36
Highlander, Tyler Clayton. « Conditional Dilated Attention Tracking Model - C-DATM ». Wright State University / OhioLINK, 2019. http://rave.ohiolink.edu/etdc/view?acc_num=wright1564652134758139.
Texte intégral
37
Sock, Ching Low. « Giving centre stage to top-down inhibitory mechanisms for selective attention ». Doctoral thesis, Universitat Pompeu Fabra, 2020. http://hdl.handle.net/10803/670753.
Texte intégralRésumé :
Selective attention determines the sensory signals that are processed at higher levels at the expense of others and is biased by higher-order brain regions which anticipate task-relevant stimuli and increase neural sensitivity to them in the sensory cortex. Often, this is thought to occur through excitation of selected neurons, but some studies have suggested that it is not the full description of the process. Increasingly, evidence has pointed to an alternative, top-down inhibitory biasing mechanism. Here, we investigated such an inhibitory model of attention. We first showed how sensitivity to stimulus features known to be task-irrelevant are reduced through top-down suppression. Secondly, we demonstrated a biologically grounded spiking model’s ability to modulate information processing and benchmarked it to physiology. Lastly, we explored the interaction between the excitatory and inhibitory models of top-down attention in a foraging agent. Our results support the inhibitory model of top-down attention as a biological attentional mechanism and show how it fits into the current zeitgeist of top-down attentional mechanisms.L’atenció selectiva determina els senyals sensorials que es processen a nivells superiors a costa dels altres. Està esbiaixada per regions cerebrals d’ordre superior que anticipen estímuls rellevants per a la tasca i augmenten la sensibilitat neuronal a l’escorça sensorial. Sovint, es creu que això es produeix mitjançant l'excitació de neurones seleccionades, però alguns estudis han suggerit que no és la descripció completa del procés. Cada vegada més, l’evidència apunta cap a un mecanisme alternatiu de polarització inhibitiva de dalt a baix. Aquí hem investigat, aleshores, un model d’atenció inhibitori. Primer, vam demostrar com es redueix la sensibilitat a les funcions d’estímul irrellevants per tasques mitjançant la supressió de dalt a baix. En segon lloc, vam demostrar la capacitat d’un model d’espiga basat en la biologia per modular el processament de la informació i l’hem comparat amb la fisiologia. Per últim, hem explorat la interacció entre els models excitadors i inhibidors d’atenció de dalt a baix en un agent de cerca d’aliments. Els nostres resultats donen suport al model inhibitori de l’atenció de dalt a baix com a mecanisme d’atenció biològica i mostren com s’adapta al ‘zeitgeist’ actual dels mecanismes d’atenció de dalt a baix.
38
Burton, Pamela Ann. « Physiological evidence of interactive object-based and space-based attention mechanisms ». Access to citation, abstract and download form provided by ProQuest Information and Learning Company ; downloadable PDF file 50.79Mb,139 p, 2005. http://wwwlib.umi.com/dissertations/fullcit/3157279.
Texte intégral
39
Cristescu, Tamara C. « Flexible mechanisms for orienting attention to words in the human brain ». Thesis, University of Oxford, 2006. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.442831.
Texte intégral
40
Austin, Alison J. « Mechanisms of attention for cues associated with rewarding and aversive outcomes ». Thesis, University of Sussex, 2010. http://sro.sussex.ac.uk/id/eprint/6265/.
Texte intégralRésumé :
Attentional biases arising from classical conditioning processes may contribute to the maintenance of drug addictions and anxiety disorders. This thesis examined whether attentional mechanisms for conditioned stimuli (CS) would be dominated by affective properties (Lang, Greenwald, Bradley, & Hamm, 1993), or the uncertainty of the stimulus in predicting the outcome (Pearce & Hall, 1980). In chapter one affective and uncertainty-driven mechanisms of attention are discussed in relation to rewarding and aversive outcomes. In experimental chapter 2 methodological issues are addressed. In experimental chapters three and four attentional mechanisms are tested using a discriminative conditioning procedure with visual stimuli of varying predictive certainty (CS+,CS+/-,CS-) for a monetary or noise outcome (US). Attention was measured using an eye-tracker, and emotional conditioning and learning were measured using Likert scales. It was found that attention was mediated by uncertainty (chapter 3), but increasing the intensity of the outcome switched attention to affective-driven mechanisms for the noise outcome (chapter 4). In a further experiment this effect on attention remained for the noise outcome even under conditions promoting uncertainty-driven mechanisms (chapter 6). When cigarettes were the unconditioned stimuli instead of money in the appetitive conditioning, attention was also mediated by stimulus affect (chapter 5). In chapter 7 the data are discussed and it is concluded that when the outcome is highly emotionally salient, affective-driven mechanisms of attention dominate over uncertainty.
41
Kulke, L. V. « Cortical mechanisms of visual attention in typically developing infants and adults ». Thesis, University College London (University of London), 2015. http://discovery.ucl.ac.uk/1471117/.
Texte intégralRésumé :
This thesis used a combined methodology of on-line eye tracking and high density EEG to study neural mechanisms of attention development in infants and adults. The extensively studied Fixation Shift Paradigm (FSP) measures the ability to shift attention between two stimuli (competition) or towards one single visible stimulus (non-competition) which improves during infancy. The novel method here overcame a number of methodological challenges to measure event-rated potentials during overt shifts of attention in competition and non-competition conditions. An experiment used eye tracking to test infants between 1 and 8 months on the FSP, establishing that this automated measure is suitable for non-verbal populations and adds precision to the developmental trends previously reported. An experiment successfully combined eye tracking and EEG to record patterns of brain activity during covert and overt attention shifts in adults. It found that neural mechanisms previously studied in covert attention shifts are similar to those in overt shifts, but differ in a frontal positivity, possibly reflecting saccade inhibition. Combined eye tracking and EEG with the original FSP showed that similar cortical mechanisms are involved in attention shifts under competition and non-competition conditions but that occipital response latencies differ at an early stage, reflecting the behavioural pattern of shorter latencies in non-competition conditions. Parallel measurements during infancy showed that the lateralisation of frontal brain responses coincides with developmental improvements in the ability to shift attention. In conclusion, the work demonstrates that, with suitable precautions taken to avoid artefacts, eye tracking and EEG can be successfully combined to monitor group-level brain mechanisms during overt attention shifts. Neurodevelopmental changes have been identified that underpin the increasing efficiency of neural attention pathways during infancy, and increased automation of responses from reliance on frontal pathways in infancy to occipital pathways in adulthood.
42
Baker, Christopher A. « Differentiating attention and motor system-based mechanisms underlying concealed knowledge detection / ». Diss., Digital Dissertations Database. Restricted to UC campuses, 2008. http://uclibs.org/PID/11984.
Texte intégral
43
Lisi, Matteo. « Mechanisms of top-down visual spatial attention : computational and behavioral investigations ». Doctoral thesis, Università degli studi di Padova, 2013. http://hdl.handle.net/11577/3423038.
Texte intégralRésumé :
This thesis examines the mechanisms underlying visual spatial attention. In particular I focused on top-‐down or voluntary attention, namely the ability to select relevant information and discard the irrelevant according to our goals. Given the limited processing resources of the human brain, which does not allow to process all the available information to the same degree, the ability to correctly allocate processing resources is fundamental for the accomplishment of most everyday tasks. The cost of misoriented attention is that we could miss some relevant information, with potentially serious consequences.
In the first study (chapter 2) I will address the issue of the neural substrates of visual spatial attention: what are the neural mechanisms that allow the deployment of visual spatial attention? According to the premotor theory orienting attention to a location in space is equivalent to planning an eye movement to the same location, an idea strongly supported by neuroimaging and neurophysiological evidence. Accordingly, in this study I will present a model that can account for several attentional effects without requiring additional mechanisms separate from the circuits that perform sensorimotor transformations for eye movements. Moreover, it includes a mechanism that allows, within the framework of the premotor theory, to explain dissociations between attention and eye movements that may be invoked to disprove it. In the second model presented (chapter 3) I will further investigate the computational mechanisms underlying sensorimotor transformations. Specifically I will show that a representation in which the amplitude of visual responses is modulated by postural signal is both efficient and plausible, emerging also in a neural network model trained through unsupervised learning (i.e., using only signals locally available at the neuron level). Ultimately this result gives additional support to the approach adopted in the first model.
Next, I will present a series of behavioral studies: in the first (chapter 4) I will show that spatial constancy of attention (i.e., the ability to sustain attention at a spatial location across eye movements) is dependent on some properties of the image, namely the presence of continuous visual landmarks at the attended locations. Importantly, this finding helps resolve contrasts between several recent results. In the second behavioral study (chapter 5), I will investigate an often neglected aspect of spatial cueing paradigms, probably the most widely used technique in studies of covert attention: the role of cue predictivity (i.e. the extent to which the spatial cue correctly indicates the location where the target stimulus will appear). Results show that, independently of participant’s awareness, changes

in predictivity result in changes in spatial validity effects, and that reliable shifts of attention can take place also in the absence of a predictive cue. In sum the results question the appropriateness of using predictive cues for delineating pure voluntary shifts of spatial attention. Finally, in the last study I will use a psychophysiological measure, the diameter of the eye’s pupil, to investigate intensive aspects of attention. Event-‐related pupil dilations accurately mirrored changes in visuospatial awareness induced by a dual-‐task manipulation that consumed attentional resources. Moreover, results of the primary spatial monitoring task revealed a significant rightward bias, indicated by a greater proportion of missed targets in the left hemifield. Interestingly this result mimics the extinction to double simultaneous stimulation (i.e., the failure to respond to a stimulus when it is presented simultaneously with another stimulus) which is often found in patients with unilateral brain damage.
Overall, these studies present an emerging picture of attention as a complex mechanism that even in its volitional aspects is modulated by other non-‐volitional factors, both external and internal to the individualQuesta tesi verte sull’indagine dei meccanismi alla base dell'attenzione visuo-‐spaziale e In particolare sull'attenzione top-‐down. Con questo termine si intende la capacità di selezionare le informazioni rilevanti e scartare quelle irrilevanti in maniera volontaria e sulla base dei nostri obiettivi. Il cervello umano non è in grado di processare allo stesso livello tutte le informazioni disponibili nell’ambiente in un dato momento, per questo una selezione corretta dell’informazione da elaborare è fondamentale anche per l’esecuzione delle più semplici attività quotidiane. Prestare attenzione ad informazioni irrilevanti può farci trascurare altre informazioni di importanza cruciale, con conseguenze potenzialmente gravi. Nel primo studio (capitolo 2) I affronterò con un approccio computazionale la questione dei meccanismi neurali che sottendono l’attenzione visuo-‐spaziale: quali sono le basi neurali dell’attenzione visuo-‐spaziale? Secondo la teoria premotoria, orientare l'attenzione verso una specifica posizione spaziale equivale a preparare un movimento oculare verso la medesima posizione, un’ipotesi supportata dai risultati di molteplici studi di neuroimaging e neurofisiologici, i quali hanno mostrato una notevole sovrapposizione tra i circuiti dedicati all’attenzione visiva e la programmazione di movimenti oculari. In questo capitolo presenterò un modello computazionale in grado di spiegare diversi effetti attentivi senza richiedere l’aggiunta di meccanismi specifici oltre ai circuiti oculomotori. Inoltre include un meccanismo, modellato sulla base di dati neurofisiologici, che consente di anticipare le conseguenze sensoriali di un movimento oculare sulla rappresentazione spaziale interna al modello, e di spiegare alcune recenti dimostrazioni di dissociazione tra attenzione e movimenti oculari che possono essere utilizzate per confutare la teoria premotoria. Nel capitolo successivo presenterò un secondo modello computazionale (capitolo 3) con lo scopo di investigare ulteriormente i meccanismi computazionali alla base delle trasformazioni sensorimotorie, cioè i processi che traducono l’informazione sensoriale in appropriati comandi motori. In particolare mostrerò che una rappresentazione spaziale costituita da neuroni con campi recettivi retinocentrici ,modulati in ampiezza da un segnale posturale, è sia efficiente (al fine di trasformare l’informazione visiva in coordinate motorie centrate su un effettore) che plausibile, in quanto emerge in un modello di rete neurale addestrato in maniera non supervisionata (usando cioè solo segnali disponibili localmente a livello  del singolo neurone). Questo risultato supporta inoltre l’approccio utilizzato nel primo modello presentato. Successivamente presenterò una serie di studi comportamentali: nel primo (capitolo 4), mostrerò che la costanza spaziale dell’attenzione visiva rispetto ai movimenti oculari (cioè la capacità di mantenere stabilmente l'attenzione in un punto nello spazio attraverso successivi movimenti oculari), dipende fortemente da alcune proprietà dell'immagine, vale a dire la presenza continua di punti di riferimento visivi. Questo risultato aiuta a risolvere recenti controversie sull’orientamento dell’attenzione durante movimenti oculari. Nel secondo studio comportamentale (capitolo 5), indagherò un aspetto spesso trascurato relativo al paradigma di cueing spaziale (probabilmente la tecnica più utilizzata nello studio dell’attenzione spaziale): la predittività del cue (cioè la misura in cui il cue spaziale indica correttamente la posizione in cui apparirà lo stimolo bersaglio). I risultati mostrano che, indipendentemente dalla consapevolezza dei partecipanti, variazioni nella predittività producono corrispondenti variazioni degli effetti di validità del cue, e che effetti significativi di validità possono comparire anche in assenza di un cue predittivo o direzionale. Questi risultati mettono in dubbio l’appropriatezza dell’uso di cue predittivi per indagare spostamenti volontari dell’attenzione spaziale. Infine, nell'ultimo studio userò una misura psicofisiologica, il diametro della pupilla, per indagare gli aspetti relativi all’intensità del processamento visuospaziale. In particolare mostrerò come dilatazioni della pupilla evento-‐relate riflettano accuratamente variazioni nella performance in un compito di monitoraggio spaziale provocate dall’aggiunta di un doppio-‐compito. Inoltre, i risultati del compito primario spaziale rivelano la presenza di un bias consistente verso l’emispazio di destra, indicato da una percentuale maggiore di bersagli omessi nell’emispazio di sinistra. In particolare il pattern di errori rispecchia il fenomeno dell’estinzione (mancata risposta a uno stimolo quando è presentata simultaneamente con un secondo stimolo, tipicamente nell’emispazio opposto) che si trova spesso in pazienti con danno cerebrale unilaterale. In conclusione, dagli studi presentati emerge un quadro dell’attenzione volontaria visuo-‐spaziale come un meccanismo complesso, che, anche nei suoi aspetti volitivi è fortemente influenzato da altri fattori, non volitivi, sia esterni che interni all'individuo
44
Wallis, George J. « Selection mechanisms for working memory ». Thesis, University of Oxford, 2014. http://ora.ox.ac.uk/objects/uuid:1a9be8d6-9f8e-49af-8185-8be4c890010f.
Texte intégralRésumé :
The experiments in this thesis investigated the mechanisms controlling input and output gating of working memory. In chapter 3, accuracy and reaction time data from a precision/capacity working memory task with prospective and retrospective cues were analysed. The results suggest that retrocues boost performance by facilitating output gating from working memory. In chapter 4, the role of perceptual cortex in mediating the cue benefits in this task was investigated with magnetoencephalography (MEG). The pattern of alpha (8-12Hz) power in visual cortex was modulated by cue direction following both precues and retrocues, but whilst this modulation was sustained following a precue (until presentation of the memory array) it was transient following a retrocue, suggesting that a memory representation was briefly retrieved or refreshed, but that there was not a sustained biasing of top-down input to visual cortex following retrocues. This argues against the standard model of working memory as sustained attention to internal representations, and in favour of a more dynamic view in which perceptual cortex is recruited transiently, and otherwise freed up for on-going processing. In chapter 5, the role of frontal networks in precueing and retrocueing was investigated. An fMRI meta-analysis identified control networks involved in preparatory and mnemonic selection: whilst the fronto-parietal network is recruited in both cases, the cingulo-opercular network is recruited only by retrocues. This spatial pattern was replicated with a source-space ROI analysis of MEG induced-responses. These data also characterised the time-course of control network activation shedding light on their functional roles. The fronto-parietal network was activated immediately following both precues and retrocues, consistent with a direct role in top-down influence over perceptual cortex. By contrast, the cingulo-opercular network was activated following retrocues only after the perceptual refreshing event was complete, suggesting a downstream role, perhaps in selecting representations to guide action. Chapter 6 investigated the role of reward associations in controlling access to working memory, testing behavioural predictions of two theories implicating the dopamine system and basal ganglia in control of working memory. The results supported a temporal gating account in which encountering reward associated items triggers a brief (<300ms) window in which there is a boost of encoding for WM. Chapter 7 discusses the implications of the current work and suggests some future directions.
45
Koleszar, Thomas W. « The generation of IPDP micropulsations, with special attention to frequency shift mechanisms ». Thesis, University of British Columbia, 1988. http://hdl.handle.net/2429/29132.
Texte intégralRésumé :
Short period geomagnetic micropulsations termed IPDPs (Intervals of Pulsations of Diminishing Period) are investigated using ground station data, geosynchronous satellite magnetograms, and the Kp and Dst geomagnetic indices. A model for the generation of IPDPs is described, and consideration is given to three mechanisms which could be responsible for the IPDP frequency rise: the inward motion, azimuthal drift, and increasing background magnetic field mechanisms. A simplified IPDP generation model containing the first two of these mechanisms is tested by computer simulation. Results from this simulation indicate the possibility of significant source region inward motion without actual plasmapause displacement, and the possibility of eastward developing IPDPs. Using amplitude variations along a north-south line of ground stations, two methods, each applicable under different ionospheric propagation conditions, are developed for quantitatively determining the inward motion of the IPDP source region. A system for qualitatively determining the potential influence of the increasing background field mechanism on an IPDP using the Dst index and geosynchronous satellite magnetograms is also formulated. Lastly, a technique for the assessment of the effects of the azimuthal drift mechanism, in conjunction with the inward motion mechanism, is developed. This technique assumes that only these two mechanisms are operating. In addition to addressing the frequency shift mechanisms, it provides estimates of the injection boundary position and the magnitude of any (ring current created) magnetic field depression in the IPDP source region. The frequency rises of two IPDPs are analyzed in detail using these methods. In both cases, the inward motion effect is the dominant factor in producing the frequency rise, with the increasing background field mechanism having no significant effect. The azimuthal drift mechanism is a secondary factor in creating one event's frequency rise, and actually suppresses the frequency rise of the other event. The computer simulation calculations also generally show the inward motion mechanism to be the dominant effect in producing IPDP frequency rises. Longitudinal variations within an IPDP event are also examined. The results of this examination are consistent with the IPDP generation model used here, which includes showing significant variations between stations spaced comparatively closely in longitude.Science, Faculty of
Earth, Ocean and Atmospheric Sciences, Department of
Graduate
46
West, Vicki Hopfinger Joseph B. « Comparisons of the neural mechanisms of voluntary, reflexive, and socially-directed attention ». Chapel Hill, N.C. : University of North Carolina at Chapel Hill, 2007. http://dc.lib.unc.edu/u?/etd,747.
Texte intégralRésumé :
Thesis (Ph. D.)--University of North Carolina at Chapel Hill, 2007.Title from electronic title page (viewed Dec. 18, 2007). "... in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Psychology." Discipline: Psychology; Department/School: Psychology.
47
Carmel, David Podhorzer. « Top-down control of visual attention and awareness : cognitive and neural mechanisms ». Thesis, University College London (University of London), 2007. http://discovery.ucl.ac.uk/1444603/.
Texte intégralRésumé :
Recent behavioural and neural research suggests that awareness is intimately related to top-down cognitive functions such as attention. Here I present a characterization of this relationship, guided by Lavie's load theory. Load theory proposes that perception has limited capacity but proceeds automatically on all stimuli (whether relevant to the task at hand or not) until capacity is exhausted, and that the allocation of processing resources to certain stimuli (rather than to other, competing ones) is guided by executive control functions such as working memory. The theory predicts that increasing the perceptual load of a task will consume capacity, therefore reducing processing of stimuli external to that task it also predicts that increasing working memory load will impair executive control, leading to increased processing of salient ignored stimuli. Here I show that these predictions hold not only for indirect measures of perceptual processing, as has been demonstrated previously, but also for visual awareness - the subjective experience of seeing and being able to report the nature of a visual stimulus. I find that under high perceptual load, observers become less aware of the very presence of other stimuli, even when these stimuli are fully expected and serve as targets. I also show that perceptual load affects the temporal resolution of visual awareness - under high load, the ability to detect a temporal pattern (luminance flicker) is reduced, leading to a subjective percept of steady illumination. In a neuroimaging study, I show that subjective awareness of flicker is associated with activity in frontal and parietal brain regions previously associated with attention and awareness. Next, I investigate the role of executive control in visual awareness by examining the effect of working memory load on binocular rivalry, a fundamental form of visual competition. I find that high working memory load reduces dominance durations in rivalry, suggesting that working memory may serve to maintain perceptual biases during competitive interactions in visual awareness. Finally, I use Transcranial Magnetic Stimulation to establish a causal role for the previously described right parietal involvement in the control of binocular rivalry. This research therefore indicates that top- down cognitive and neural mechanisms are involved in determining whether visual stimuli will reach awareness, and in shaping the subjective nature of the experience such stimuli evoke.
48
MacNamara, Kailey. « Behavioral and Neural Mechanisms of Social Heterogeneity in Attention Deficit/Hyperactivity Disorder ». FIU Digital Commons, 2017. http://digitalcommons.fiu.edu/etd/3390.
Texte intégralRésumé :
Attention-deficit/hyperactivity disorder (ADHD) is one of the most common child-onset neurodevelopment disorders, affecting 5% of children in the United States (American Psychiatric Association, 2013). Treatment matching in ADHD is difficult and unsatisfactory; the same general treatment algorithm is recommended for all children. It is therefore important to consider the development of specialized treatment programs based on a variety of behavioral and neurological biomarkers. Unfortunately, due to its multi-faceted classification, the heterogeneity of this behavioral disorder is under-investigated (Costa Dias et al., 2015). Scientific research in this area is especially limited as the severity of ADHD goes undiagnosed, children tend to have difficulties remaining still in MRI scanners, and the hyperactivity-impulsivity that is associated with ADHD may cause further challenges when trying to remain motionless in the scanner. Furthermore, tasks such as Facial Emotion Perception Task (FEPT) and Theory of Mind (ToM) have not been used to analyze social and behavioral deficits in children with ADHD. More research needs to be allocated to helping uncover the neural substrates underlying the inattention and hyperactivity traits of this disorder. For this reason, we acquired functional magnetic resonance imaging (fMRI) data from five children with ADHD performing the FEPT and ToM tasks. The results showed the children have an easier and quick time correctly identifying happy emotional states, as compared to the fearful, angry, and neutral conditions. Results from the FEPT task also revealed that the participants were thinking and reasoning more (i.e., taking longer to deduce an ending) when identifying emotions than identifying animals. The ToM task showed that the default mode network (DMN) may not be fully suppressed when the children are choosing the correct cartoon ending, and therefore the children may be having lapses in attention. These findings may assist the current hypothesis that the default mode network has reduced network homogeneity in people with ADHD. Overall, the findings presented in this thesis provide a good diving board into discovering the reason(s) for the social cognition and emotion recognition impairments associated with ADHD, but further research is needed in order to one day pinpoint and ultimately correct the regions(s) of dysfunction.
49
Walters-Symons, Rosanna Mary. « Attention, gaze, response programming : examining the cognitive mechanisms underpinning the quiet eye ». Thesis, University of Exeter, 2017. http://hdl.handle.net/10871/28336.
Texte intégralRésumé :
The quiet eye (QE) - the final fixation or tracking gaze on a specific location that has an onset prior to the start of a final, critical movement (Vickers, 2007) - has emerged as a key predictor of proficient performance in targeting and interceptive tasks over the last 20 years. Since Vickers’ seminal study in golf putting, the QE has been examined in over 28 different motor tasks, with a longer QE duration often referred to as a characteristic of superior performance and a measure of optimal visuo-motor control. However, the underpinnings of QE are not fully understood, with many researchers advocating the need to better identify and understand the mechanisms that underlie the QE (Williams, 2016; Gonzalez et al., 2015). Consequently, the overriding goal of this thesis was to examine the function of the QE duration, what it represents and how it exerts an influence, by exploring the attentional underpinnings of the QE and the prominent cognitive mechanism of response programming. In study 1 (chapter 4), the manipulation of different parameters of golf putting and the examination of different response programming functions (pre-programming vs online control) during the QE enabled me to build on previous explorations of the response programming function by investigating QE’s response to specific iterations of increased task demands. Experienced golfers revealed that longer QE durations were found for more complex iterations of the task and more sensitive analyses of the QE proportions suggest that the early QE (prior to movement initiation) is closely related to force production and impact quality. While the increases in QE were not functional in terms of supporting improved performance, the longer QE durations may have had a positive, insulating effect. In study 2 (chapter 5), a re-examination of Vickers’ seminal work in golf putting was performed, taking into account an error recovery perspective. This 3 explored the influence of trial-to-trial dependence on the functionality of the QE duration and the possible compensatory mechanism that assists in the re-parameterisation of putting mechanics following an unsuccessful trial. The results reveal that experienced golfers had consistently longer QE durations than novices but there was no difference in QE between randomly chosen hits and misses. However, QE durations were significantly longer on hits following a miss, reflecting a potential error recovery mechanism. Importantly, QE durations were significantly lower on misses following a miss, suggesting that motivation moderates the adoption of a compensatory longer QE strategy. These findings indicate that the QE is influenced by the allocation of attentional effort. To explore this notion further, in study 3 (chapter 6), two experiments were undertaken. Experiment 1 examined the QE’s response to attentional effort that is activated via goal motivation and experiment 2 examined the effect of disrupting the allocation of attentional effort on the QE using a dual-task paradigm. The early proportion of the QE was sensitive to motivation, indicating that the QE is not purely determined by the demands of the task and golfers have the ability to apply attentional effort, and therefore QE, strategically (exp. 1). The results also support the assumption that QE reflects overt attentional control but question the sensitivity of QE to detect movements in the locus of attentional effort that does not activate shifts in gaze (covert attention) (exp. 2). The results in this thesis conclude that, while significant contributions to understanding what the QE represents and how it may exert its influence are made, there still remains unanswered questions and tensions that require exploration.
50
Magimairaj, Beula M. « Attentional mechanisms in children's complex memory span performance ». Ohio : Ohio University, 2010. http://www.ohiolink.edu/etd/view.cgi?ohiou1267650640.
Texte intégral