Mechanisms Of Selective Attention Research Articles

An Automated Approach for Predicting HAMD-17 Scores via Divergent Selective Focused Multi-heads Self-Attention Network

This study introduces the Divergent Selective Focused Multi-heads Self-Attention Network (DSFMANet), an innovative deep learning model devised to automatically predict Hamilton Depression Rating Scale-17 (HAMD-17) scores in patients with depression. This model introduces a multi-branch structure for sub-bands and artificially configures attention focus factors on various branches, resulting in distinct attention distributions for different sub-bands. Experimental results demonstrate that when DSFMANet processes sub-band data, its performance surpasses current benchmarks in key metrics such as mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R2). This success is particularly evident in terms of MSE and MAE, where the performance of sub-band data is significantly superior, highlighting the model's potential in accurately predicting HAMD-17 scores. Concurrently, the experiment also compared the model's performance with sub-band and full-band data, affirming the superiority of the selective focus attention mechanism in electroencephalography (EEG) signal processing. DSFMANet, when utilizing sub-band data, exhibits higher data processing efficiency and reduced model complexity. The findings of this study underscore the significance of employing deep learning models based on sub-band analysis in depression diagnosis. The DSFMANet model not only effectively enhances the accuracy of depression diagnosis but also offers valuable research directions for similar applications in the future. This deep learning-based automated approach can effectively ascertain the HAMD-17 scores of patients with depression, thus offering more accurate and reliable support for clinical decision-making.

Selective social interactions and speed-induced leadership in schooling fish

Animals moving together in groups are believed to interact among each other with effective social forces, such as attraction, repulsion, and alignment. Such forces can be inferred using "force maps," i.e., by analyzing the dependency of the acceleration of a focal individual on relevant variables. Here, we introduce a force map technique suitable for the analysis of the alignment forces experienced by individuals. After validating it using an agent-based model, we apply the force map to experimental data of schooling fish. We observe signatures of an effective alignment force with faster neighbors and an unexpected antialignment with slower neighbors. Instead of an explicit antialignment behavior, we suggest that the observed pattern is the result of a selective attention mechanism, where fish pay less attention to slower neighbors. This mechanism implies the existence of temporal leadership interactions based on relative speeds between neighbors. We present support for this hypothesis both from agent-based modeling as well as from exploring leader-follower relationships in the experimental data.

Generalized anxiety disorder and selective attention: An unsuccessful replication of Yiend et al., (2015) in a student population

ABSTRACT Generalized Anxiety Disorder (GAD) is an anxiety disorder that is believed to affect attention (Stein, M. B., & Sareen, J. (2015). Generalized anxiety disorder. New England Journal of Medicine, 373(21), 2059–2068. https://doi.org/10.1056/NEJMcp1502514; Yiend, J., Mathews, A., Burns, T., Dutton, K., Fernández-Martín, A., Georgiou, G. A., Luckie, M., Rose, A., Russo, R., & Fox, E. (2015). Mechanisms of selective attention in generalized anxiety disorder. Clinical Psychological Science, 3(5), 758–771. https://doi.org/10.1177/2167702614545216). Previous literature has found that selective attention is changed when someone perceives threatening stimuli, such as an angry face, and that those with anxiety disorders, may have a heightened or delayed response to threatening stimuli (Richards, H. J., Benson, V., Donnelly, N., & Hadwin, J. A. (2014). Exploring the function of selective attention and hypervigilance for threat in anxiety. Clinical Psychology Review, 34(1), 1–13. https://doi.org/10.1016/j.cpr.2013.10.006; Stevens, C., & Bavelier, D. (2012). The role of selective attention on academic foundations: A cognitive neuroscience perspective. Developmental Cognitive Neuroscience, 2, S30–S48. https://doi.org/10.1016/j.dcn.2011.11.001), which may alter how fast a presented task is completed (Yiend et al., 2015). The present study aimed to reproduce findings by Yiend et al. (2015), which identified an unexpected pattern in those with GAD: faster disengagement from angry faces compared to positive (happy, neutral) faces. The present study recruited a larger (nonclinical) sample from a student population to achieve greater statistical power. None of the findings reported by Yiend and colleagues (Experiment 1; 2015) were replicated in a student sample. The implications are discussed.

Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing

GAN-based image attribute editing firstly leverages GAN Inversion to project real images into the latent space of GAN and then manipulates corresponding latent codes. Recent inversion methods mainly utilize additional high-bit features to improve image details preservation, as low-bit codes cannot faithfully reconstruct source images, leading to the loss of details. However, during editing, existing works fail to accurately complement the lost details and suffer from poor editability. The main reason is they inject all the lost details indiscriminately at one time, which inherently induces the position and quantity of details to overfit source images, resulting in inconsistent content and artifacts in edited images. This work argues that details should be gradually injected into both the reconstruction and editing process in a multi-stage coarse-to-fine manner for better detail preservation and high editability. Therefore, a novel dual-stream framework is proposed to accurately complement details at each stage. The Reconstruction Stream is employed to embed coarse-to-fine lost details into residual features and then adaptively add them to the GAN generator. In the Editing Stream, residual features are accurately aligned by our Selective Attention mechanism and then injected into the editing process in a multi-stage manner. Extensive experiments have shown the superiority of our framework in both reconstruction accuracy and editing quality compared with existing methods.

A novel feature-level fusion scheme with multimodal attention CNN for heart sound classification

Background and objectiveMost of the existing machine learning-based heart sound classification methods achieve limited accuracy. Since they primarily depend on single domain feature information and tend to focus equally on each part of the signal rather than employing a selective attention mechanism. In addition, they fail to exploit convolutional neural network (CNN) - based features with an effective fusion strategy. MethodsIn order to overcome these limitations, a novel multimodal attention convolutional neural network (MACNN) with a feature-level fusion strategy, in which Mel-cepstral domain as well as general frequency domain features are incorporated to increase the diversity of the features, is proposed in this paper. In the proposed method, DilationAttenNet is first utilized to construct attention-based CNN feature extractors and then these feature extractors are jointly optimized in MACNN at the feature-level. The attention mechanism aims to suppress irrelevant information and focus on crucial diverse features extracted from the CNN. ResultsExtensive experiments are carried out to study the efficacy of the feature level fusion in comparison to that with early fusion. The results show that the proposed MACNN method significantly outperforms the state-of-the-art approaches in terms of accuracy and score for the two publicly available Github and Physionet datasets. ConclusionThe findings of our experiments demonstrated the high performance for heart sound classification based on the proposed MACNN, and hence have potential clinical usefulness in the identification of heart diseases. This technique can assist cardiologists and researchers in the design and development of heart sound classification methods.

A mechanism for deviance detection and contextual routing in the thalamus: a review and theoretical proposal.

Predictive processing theories conceptualize neocortical feedback as conveying expectations and contextual attention signals derived from internal cortical models, playing an essential role in the perception and interpretation of sensory information. However, few predictive processing frameworks outline concrete mechanistic roles for the corticothalamic (CT) feedback from layer 6 (L6), despite the fact that the number of CT axons is an order of magnitude greater than that of feedforward thalamocortical (TC) axons. Here we review the functional architecture of CT circuits and propose a mechanism through which L6 could regulate thalamic firing modes (burst, tonic) to detect unexpected inputs. Using simulations in a model of a TC cell, we show how the CT feedback could support prediction-based input discrimination in TC cells by promoting burst firing. This type of CT control can enable the thalamic circuit to implement spatial and context selective attention mechanisms. The proposed mechanism generates specific experimentally testable hypotheses. We suggest that the L6 CT feedback allows the thalamus to detect deviance from predictions of internal cortical models, thereby supporting contextual attention and routing operations, a far more powerful role than traditionally assumed.

Phasic alerting in visual search tasks.

Many tasks require one to search for and find important objects in the visual environment. Visual search is strongly supported by cues indicating target objects to mechanisms of selective attention, which enable one to prioritise targets and ignore distractor objects. Besides selective attention, a major influence on performance across cognitive tasks is phasic alertness, a temporary increase of arousal induced by warning stimuli (alerting cues). Alerting cues provide no specific information on whose basis selective attention could be deployed, but have nevertheless been found to speed up perception and simple actions. It is still unclear, however, how alerting affects visual search. Therefore, in the present study, participants performed a visual search task with and without preceding visual alerting cues. Participants had to report the orientation of a target among several distractors. The target saliency was low in Experiment 1 and high in Experiment 2. In both experiments, we found that visual search was faster when a visual alerting cue was presented before the target display. Performance benefits occurred irrespective of how many distractors had been presented along with the target. Taken together, the findings reveal that visual alerting supports visual search independently of the complexity of the search process and the demands for selective attention.

Interaction-and-Response Network for Distantly Supervised Relation Extraction.

Distantly supervised relation extraction (DSRE) aims to identify semantic relations from massive plain texts. A broad range of the prior research has leveraged a series of selective attention mechanisms over sentences in a bag to extract relation features without considering dependencies among the relation features. As a result, potential discriminative information existed in the dependencies is ignored, causing a decline in the performance of extracting entity relations. In this article, we focus on going beyond the selective attention mechanisms and propose a new framework termed interaction-and-response network (IR-Net) that adaptively recalibrates the features of sentence, bag, and group levels by explicitly modeling interdependencies among the features on each level. The IR-Net consists of a series of interactive and responsive modules throughout feature hierarchy, seeking to strengthen its power of learning salient discriminative features for distinguishing entity relations. We conduct extensive experiments on three benchmark DSRE datasets, including NYT-10, NYT-16, and Wiki-20m. The experimental results demonstrate that the IR-Net brings obvious improvements in performance when comparing ten state-of-the-art DSRE methods for entity relation extraction.

FACIAL EMOTION RECOGNITION BASED ON SELECTIVE KERNEL NETWORK

Existing deep learning methods for facial emotion recognition only focus on optimizing network structures, utilizing fixed receptive fields for different images, and relying on feature extraction based on a single scale of receptive fields. However, this approach fails to fully capture the most critical facial regions. To address this limitation, this paper presents a novel technique for facial emotion recognition that employs a selective kernel network. The proposed method introduces a dedicated module called the selective kernel network, which is trained using transfer learning. This module incorporates various components, such as a selective attention mechanism and channel-wise independent feature extraction and fusion. These components allow for the extraction of feature information from key facial regions. Unlike other methods, the selective convolutional kernel network extracts features with multiple scales of receptive fields and adapts to different spatial positions using a multilayer perceptron. This adaptability enhances useful features and suppresses noise. After extracting the features, they are combined, and the classification outcome is computed using the softmax function. Experimental results demonstrate that the suggested approach achieves an accuracy of 88.4 and 92.1% on the RAF-DB and KDEF datasets, respectively. These results confirm the efficacy of the proposed technique in comprehensively capturing the most crucial facial regions. Moreover, compared to alternative methods, this technique exhibits superior accuracy and enhanced resilience.

Empowering Semantic Segmentation with Selective Frequency Enhancement and Attention Mechanism for Tampering Detection

Empowering Semantic Segmentation with Selective Frequency Enhancement and Attention Mechanism for Tampering Detection

Facial symmetry perception and attractiveness ratings in body dysmorphic disorder

Cognitive-behavioral models of body dysmorphic disorder (BDD) conceptualize biases in visual information processing as maintaining factors of the disorder. Empirical findings point to selective attentional mechanisms and a detail-oriented information processing style in facial perception, which are related to perceived facial aesthetics. However, so far, it remains unclear whether higher discrimination abilities of facial symmetry and related attractiveness ratings contribute to the mind as a filter in BDD. BDD (n = 25) and mentally healthy subjects (n = 25) rated symmetry and attractiveness of fifty animated, three-dimensional facial photographs characterized by diverging geometrical symmetry. Main findings emerging from mixed effects models indicated that perceived symmetry, but not objective symmetry, predicted perceived attractiveness. Group status did not predict symmetry or attractiveness ratings. Exploratory analyses indicated potentially clinically relevant mechanisms of symmetry ratings in BDD and revealed faster attractiveness ratings in unattractive vs. attractive faces across groups. Our findings refute the hypothesis of higher discrimination abilities of facial symmetry as a maintaining factor in BDD. Implications for attention and perception retraining in the context of cognitive-behavioral therapy for BDD are discussed.

Contingency learning on the proportional congruency effect of the Stroop task is manifest in deviant processing.

The proportion congruency effect has been reported in interference tasks. The mostly congruent block is the frequent condition, and the mostly incongruent block is the rare condition for congruent stimuli, whereas the mostly congruent block is the rare condition, and the mostly incongruent block is the frequent condition for incongruent stimuli. This study examined the cognitive mechanism underlying the proportion congruency effect in the two-choice Stroop task using event-related potential components related to deviant processing. Stimuli were Kanji characters meaning 'red' and 'blue' painted by congruent and incongruent colors. Participants were required to provide a two-choice button-press response corresponding to the colors. The congruent and incongruent stimuli were presented on rare (20%) and frequent (80%) conditions. N1 was enhanced in the rare condition relative to the frequent condition for both congruent and incongruent stimuli. The results suggested that colors and characters were not processed independently in the N1 time range, which made selective attention to the relevant feature difficult. Posterior negativity from 200 to 250 ms was also different between rare and frequent conditions, indicating the presence of visual mismatch negativity for congruent and incongruent stimuli. It was considered that the distinction between congruent and incongruent stimuli was evident in mostly incongruent blocks, indicating that the selective attention mechanism was not actively engaged. The proportion congruency effect was explained by contingency learning rather than selective attention in the present task. The cognitive mechanisms underlying the proportion congruency effect are reflected by deviant event-related potential components.

Forward Collision Warning Strategy Based on Millimeter-Wave Radar and Visual Fusion

Forward collision warning (FCW) is a critical technology to improve road safety and reduce traffic accidents. However, the existing multi-sensor fusion methods for FCW suffer from a high false alarm rate and missed alarm rate in complex weather and road environments. For these issues, this paper proposes a decision-level fusion collision warning strategy. The vision algorithm and radar tracking algorithm are improved in order to reduce the false alarm rate and omission rate of forward collision warning. Firstly, this paper proposes an information entropy-based memory index for an adaptive Kalman filter for radar target tracking that can adaptively adjust the noise model in a variety of complex environments. Then, for visual detection, the YOLOv5s model is enhanced in conjunction with the SKBAM (Selective Kernel and Bottleneck Attention Mechanism) designed in this paper to improve the accuracy of vehicle target detection. Finally, a decision-level fusion warning fusion strategy for millimeter-wave radar and vision fusion is proposed. The strategy effectively fuses the detection results of radar and vision and employs a minimum safe distance model to determine the potential danger ahead. Experiments are conducted under various weather and road conditions, and the experimental results show that the proposed algorithm reduces the false alarm rate by 11.619% and the missed alarm rate by 15.672% compared with the traditional algorithm.

Point Cloud Completion Via Skeleton-Detail Transformer.

Point cloud shape completion plays a central role in diverse 3D vision and robotics applications. Early methods used to generate global shapes without local detail refinement. Current methods tend to leverage local features to preserve the observed geometric details. However, they usually adopt the convolutional architecture over the incomplete point cloud to extract local features to restore the diverse information of both latent shape skeleton and geometric details, where long-distance correlation among the skeleton and details is ignored. In this work, we present a coarse-to-fine completion framework, which makes full use of both neighboring and long-distance region cues for point cloud completion. Our network leverages a Skeleton-Detail Transformer, which contains cross-attention and self-attention layers, to fully explore the correlation from local patterns to global shape and utilize it to enhance the overall skeleton. Also, we propose a selective attention mechanism to save memory usage in the attention process without significantly affecting performance. We conduct extensive experiments on the ShapeNet dataset and real-scanned datasets. Qualitative and quantitative evaluations demonstrate that our proposed network outperforms current state-of-the-art methods.

DSKCA-UNet: Dynamic selective kernel channel attention for medical image segmentation.

U-Net has attained immense popularity owing to its performance in medical image segmentation. However, it cannot be modeled explicitly over remote dependencies. By contrast, the transformer can effectively capture remote dependencies by leveraging the self-attention (SA) of the encoder. Although SA, an important characteristic of the transformer, can find correlations between them based on the original data, secondary computational complexity might retard the processing rate of high-dimensional data (such as medical images). Furthermore, SA is limited because the correlation between samples is overlooked; thus, there is considerable scope for improvement. To this end, based on Swin-UNet, we introduce a dynamic selective attention mechanism for the convolution kernels. The weight of each convolution kernel is calculated to fuse the results dynamically. This attention mechanism permits each neuron to adaptively modify its receptive field size in response to multiscale input information. A local cross-channel interaction strategy without dimensionality reduction was introduced, which effectively eliminated the influence of downscaling on learning channel attention. Through suitable cross-channel interactions, model complexity can be significantly reduced while maintaining its performance. Subsequently, the global interaction between the encoder features is used to extract more fine-grained features. Simultaneously, the mixed loss function of the weighted cross-entropy loss and Dice loss is used to alleviate category imbalances and achieve better results when the sample number is unbalanced. We evaluated our proposed method on abdominal multiorgan segmentation and cardiac segmentation datasets, achieving Dice similarity coefficient and 95% Hausdorff distance metrics of 80.30 and 14.55%, respectively, on the Synapse dataset and Dice similarity coefficient metrics of 90.80 on the ACDC dataset. The experimental results show that our proposed method has good generalization ability and robustness, and it is a powerful tool for medical image segmentation.

FFSDF: An improved fast face shadow detection framework based on channel spatial attention enhancement

Recent shadow detectors excel on simple datasets but encounter difficulties with facial shadow images under complex lighting due to the lack of annotated shadow masks, varying shadow sizes, and imbalances between the target and background. This leads to training difficulties, reduced accuracy, and slower processing, posing a significant challenge for precise and fast detection framework development. We collected images and created Extended Yale Shadow Detection Dataset (EYSDD). In comparison to other datasets, this dataset includes additional manually annotated shadow masks, making it suitable for training convolutional neural networks. To address this problem, we propose incorporating Channel Spatial Direction-aware Spatial Context (CSDSC) module into Fast Shadow Detection Network (FSDNet). Additionally, we introduce Selective Attention Inverted Residual Bottleneck (SAIRB) with Selective Attention Mechanism (SAM). Furthermore, we integrate Detail Enhancement Module (DEM), which refines low-level features, into Fast Face Shadow Detection Framework (FFSDF). Finally, compared to other methods, our model surpasses the baseline method FSDNet and the advanced method EVP by 3.5% and 1.9% in terms of IoU, and 1.8% and 4.3% in terms of Dice score, respectively. Our model has only 4.31 M parameters and achieves a computing speed of 0.022 sec/image, demonstrating superior efficiency compared to other methods.

SA-VAE: a novel approach for reservoir characterization based on variational auto-encoder and selective attention mechanism

SA-VAE: a novel approach for reservoir characterization based on variational auto-encoder and selective attention mechanism

Dual-band polarimetric HRRP recognition via a brain-inspired multi-channel fusion feature extraction network.

Radar high-resolution range profile (HRRP) provides geometric and structural information of target, which is important for radar automatic target recognition (RATR). However, due to the limited information dimension of HRRP, achieving accurate target recognition is challenging in applications. In recent years, with the rapid development of radar components and signal processing technology, the acquisition and use of target multi-frequency and polarization scattering information has become a significant way to improve target recognition performance. Meanwhile, deep learning inspired by the human brain has shown great promise in pattern recognition applications. In this paper, a Multi-channel Fusion Feature Extraction Network (MFFE-Net) inspired by the human brain is proposed for dual-band polarimetric HRRP, aiming at addressing the challenges faced in HRRP target recognition. In the proposed network, inspired by the human brain's multi-dimensional information interaction, the similarity and difference features of dual-frequency HRRP are first extracted to realize the interactive fusion of frequency features. Then, inspired by the human brain's selective attention mechanism, the interactive weights are obtained for multi-polarization features and multi-scale representation, enabling feature aggregation and multi-scale fusion. Finally, inspired by the human brain's hierarchical learning mechanism, the layer-by-layer feature extraction and fusion with residual connections are designed to enhance the separability of features. Experiments on simulated and measured datasets verify the accurate recognition capability of MFFE-Net, and ablative studies are conducted to confirm the effectiveness of components of network for recognition.

The time course of category-based attentional template pre-activation depends on the category framework

When searching for a target defined by a set of objects, attention can be directed toward task-relevant objects by creating a category-based attentional template (CAT). Previous studies have found that CAT can be activated before the onset of the target. However, the time course of CAT pre-activation and whether the category framework (prototypical or semantic) can modulate it remain unclear. To explore the time course of CAT pre-activation, we employed a rapid serial probe presentation paradigm (RSPP) with event-related potentials (ERPs). To investigate the effect of the category framework on the time course of CAT pre-activation, the target category was defined as the prototypical category (Experiment 1) or the semantic category (Experiment 2). The results showed that the prototype-based CAT was pre-activated 300 ms prior to the target, whereas the semantics-based CAT was pre-activated 1500 ms before the onset of the target. The difference in the time course of pre-activation between the two CAT types indicates that the category framework can modulate the time course of CAT pre-activation. Additionally, during the attentional selection phase, an overall comparison of the target revealed that a larger N2pc was elicited by the prototype-based CAT than by the semantics-based CAT, suggesting that the prototype-based CAT could capture more attention than the semantics-based CAT. The findings on the difference between the two CAT frameworks in the preparatory and attentional selection phases provide more evidence for categorical information in visual search and extend our understanding of the mechanism of categorical attentional selection.

Top-down control of exogenous attentional selection is mediated by beta coherence in prefrontal cortex

Salience-driven exogenous and goal-driven endogenous attentional selection are two distinct forms of attention that guide selection of task-irrelevant and task-relevant targets in primates. Top-down attentional control mechanisms enable selection of the task-relevant target by limiting the influence of sensory information. Although the lateral prefrontal cortex (LPFC) is known to mediate top-down control, the neuronal mechanisms of top-down control of attentional selection are poorly understood. Here, we trained two rhesus monkeys on a two-target, free-choice luminance-reward selection task. We demonstrate that visual-movement (VM) neurons and nonvisual neurons or movement neurons encode exogenous and endogenous selection. We then show that coherent beta activity selectively modulates mechanisms of exogenous selection specifically during conflict and consequently may support top-down control. These results reveal the VM-neuron-specific network mechanisms of attentional selection and suggest a functional role for beta-frequency coherent neural dynamics in the modulation of sensory communication channels for the top-down control of attentional selection.