Sequential Modules Research Articles

STQD-Det: Spatio-Temporal Quantum Diffusion Model for Real-Time Coronary Stenosis Detection in X-Ray Angiography.

Detecting coronary stenosis accurately in X-ray angiography (XRA) is important for diagnosing and treating coronary artery disease (CAD). However, challenges arise from factors like breathing and heart motion, poor imaging quality, and the complex vascular structures, making it difficult to identify stenosis fast and precisely. In this study, we proposed a Quantum Diffusion Model with Spatio-Temporal Feature Sharing to Real-time detect Stenosis (STQD-Det). Our framework consists of two modules: Sequential Quantum Noise Boxes module and spatio-temporal feature module. To evaluate the effectiveness of the method, we conducted a 4-fold cross-validation using a dataset consisting of 233 XRA sequences. Our approach achieved the F1 score of 92.39% with a real-time processing speed of 25.08 frames per second. These results outperform 17 state-of-the-art methods. The experimental results show that the proposed method can accomplish the stenosis detection quickly and accurately.

Effect of an online training intervention on evidence-based practice in clinical nurses. #Evidencer Project.

Online learning is becoming increasingly essential for health professionals, and it is necessary to understand how this modality affects clinical nurses' learning of Evidence-Based Practice (EBP). For this reason, the present study sought to assess the effectiveness of an online training program in improving nurses' EBP competence. A quasi-experimental study with a pretest-posttest design was conducted with a control group and without randomization for a period of 6months. The intervention was an online self-learning course based on Melnyk's seven steps of Evidence-Based Practice (EBP). The course was structured into sequential modules requiring 72h of work, with task completions and tests necessary for progression. Participants had three months to complete the course. EBP competence was measured with the EBP-COQ Prof© questionnaire. Data analysis included percentages, means, standard deviations, chi-square tests, student's t-tests, and a two-way repeated measuresanalysis of covariance (ANCOVA). The analysis of the changes observed in each dimension and the overall EBP competence between the intervention group (IG) and the control group (CG) showed a significant group*time interaction in three of the four dimensions of the questionnaire. The results indicated that the online Evidence-Based Practice (EBP) course significantly improved knowledge, skills, and utilization three months after its completion. In the IG, the mean score was 44.04 (standard deviation (SD) = 7), compared to 37.83 (SD = 8.5) in the CG (p < 0.001). Regarding skills, the IG had a mean score of 24.24 (SD = 3.8), while the CG scored 23.01 (SD = 3.1) (p = 0.008). For utilization, the mean score in the IG was 36.77 (SD = 6.8), and the CG was 33.12 (SD = 6.3) (p = 0.005). Overall competence also showed a statistically significant difference, with the IG achieving a mean score of 141.22 (SD = 20.0) compared to the CG with a mean score of 130.34 (SD = 16.7) (p < 0.001). Training through an online education platform for three months (72h) is an effective tool for improving the competence in Evidence-Based Practice (EBP) of clinical nurses. A significant increase was observed in knowledge and moderate improvements in skills and the application of EBP. These online courses, adapted to the needs of professionals, can be an efficient way to prepare nurses and improve their application of EBP in a clinical setting.

Neuroinflammation as a cause of differential Müller cell regenerative responses to retinal injury.

Unlike mammals, some nonmammalian species recruit Müller glia for retinal regeneration after injury. Identifying the underlying mechanisms may help to foresee regenerative medicine strategies. Using a Xenopus model of retinitis pigmentosa, we found that Müller cells actively proliferate upon photoreceptor degeneration in old tadpoles but not in younger ones. Differences in the inflammatory microenvironment emerged as an explanation for such stage dependency. Functional analyses revealed that enhancing neuroinflammation is sufficient to trigger Müller cell proliferation, not only in young tadpoles but also in mice. In addition, we showed that microglia are absolutely required for the response of mouse Müller cells to mitogenic factors while negatively affecting their neurogenic potential. However, both cell cycle reentry and neurogenic gene expression are allowed when applying sequential pro- and anti-inflammatory treatments. This reveals that inflammation benefits Müller glia proliferation in both regenerative and nonregenerative vertebrates and highlights the importance of sequential inflammatory modulation to create a regenerative permissive microenvironment.

Developing a Guided Web App for Postpartum Depression Symptoms: User-Centered Design Approach.

Psychological internet-based interventions have shown promise in preventing and treating perinatal depression, but their effectiveness can be hindered by low user engagement. This challenge often arises from a misalignment between technology attributes, user needs, and context. A user-centered, iterative approach involving all stakeholders is recommended. In this paper, we aimed to develop a user-friendly psychological internet-based intervention aimed at addressing the symptoms of perinatal depression through an iterative, user-centered approach. The development process followed the Center for eHealth Research and Disease Management Roadmap phases of contextual inquiry, value specification, and design. It involved a comprehensive literature review, 2 surveys, 10 focus groups, 5 usability interviews, and 1 technical pilot. The contextual inquiry revealed a demand for accessible interventions for perinatal mental health, with internet-based solutions seen as viable options. Insights from the literature influenced intervention content and features. Stakeholders' openness to the intervention became evident during this phase, along with the integration of the first set of values. Initially, we assessed the broader perinatal context to identify the optimal period for the intervention. On the basis of the findings and practical considerations, we decided to specifically target postpartum depression symptoms. The value specification phase further defined the central values and translated them into requirements. In the design phase, feedback was obtained on the user experience of an early digital prototype and on the prototype's final version. The resulting intervention, named Mamá, te entiendo ("Mom, I get you"), is a guided web app based on cognitive behavioral therapy principles, integrating elements from attachment and mentalization theories. It aims to reduce depressive symptoms in women during the first months postpartum and consists of 6 core sequential modules, along with 3 additional modules, including 5 case examples illustrating depressive symptoms and therapeutic techniques. The intervention provides homework exercises and offers users the opportunity to receive feedback from an e-coach through the web app. This study emphasizes the importance of a user-centered and iterative development process for psychological internet-based interventions. This process helps clarify user needs and provides valuable feedback on service design and quality, ultimately having the potential to enhance the utility and, presumably, the effectiveness of the intervention. The Discussion section shares valuable insights from the project, such as the value of the requirement sessions.

HIST: Hierarchical and sequential transformer for image captioning

AbstractImage captioning aims to automatically generate a natural language description of a given image, and most state‐of‐the‐art models have adopted an encoder–decoder transformer framework. Such transformer structures, however, show two main limitations in the task of image captioning. Firstly, the traditional transformer obtains high‐level fusion features to decode while ignoring other‐level features, resulting in losses of image content. Secondly, the transformer is weak in modelling the natural order characteristics of language. To address theseissues, the authors propose a HIerarchical and Sequential Transformer (HIST) structure, which forces each layer of the encoder and decoder to focus on features of different granularities, and strengthen the sequentially semantic information. Specifically, to capture the details of different levels of features in the image, the authors combine the visual features of multiple regions and divide them into multiple levels differently. In addition, to enhance the sequential information, the sequential enhancement module in each decoder layer block extracts different levels of features for sequentially semantic extraction and expression. Extensive experiments on the public datasets MS‐COCO and Flickr30k have demonstrated the effectiveness of our proposed method, and show that the authors’ method outperforms most of previous state of the arts.

The interplay of cognitive control and feature integration: insights from theta oscillatory dynamics during conflict processing.

Adaptive behavior is fundamental to cognitive control and executive functioning. This study investigates how cognitive control mechanisms and episodic feature retrieval interact to influence adaptiveness, focusing particularly on theta (4 to 8Hz) oscillatory dynamics. We conducted two variations of the Simon task, incorporating response-incompatible, response-compatible, and neutral trials. Experiment 1 demonstrated that cognitive adjustments-specifically, cognitive shielding following incompatible trials and cognitive relaxation following compatible ones-are reflected in midfrontal theta power modulations associated with the Simon effect. Experiment 2 showed that reducing feature overlap between trials leads to less pronounced sequential modulations in behavior and midfrontal theta activity, supporting the hypothesis that cognitive control and feature integration share a common neural mechanism. These findings highlight the interaction of cognitive control processes and episodic feature integration in modulating behavior. The results advocate for hybrid models that combine top-down and bottom-up processes as a comprehensive framework to understand cognitive control dynamics and adaptive behavior.

SiSe: Simultaneous and Sequential Transformers for multi-label activity recognition

Multi-label activity recognition is extremely challenging, where multiple activities may appear simultaneously or sequentially in a video. While previous works have realized the temporal co-occurrence of activities, the sequential order of activities have been largely overlooked. However, we argue that the sequential order of activities should also be preserved in correlation modeling, because shuffling the order might not form a semantically meaningful video. In this work, we present plug-and-play Simultaneous and Sequential Transformer (SiSe) modules for multi-label activity recognition. Upon frame features of all time steps, SiSe enhances spatio-temporal feature learning for multi-label activity recognition, by capturing the simultaneous and sequential activity correlations. Specifically, we employ a Simultaneous Transformer module to connect multiple activities that probably appear at each frame, and a hierarchical Sequential Transformer module to efficiently capture the sequential activity correlations in an order-preserved manner. Despite the straightforward and class-agnostic design of SiSe, it can outperform state-of-the-art approaches on three multi-label activity recognition benchmarks. In particular, we verify the significance of preserving the sequential order of activities with our Sequential Transformer in correlation modeling. We also conduct ablation studies and visual analysis for better understanding of our SiSe.

Multi-branch progressive embedding network for crowd counting

Crowd counting is essential for video surveillance and public safety. The performance of counting models has been greatly improved with the rapid development of Convolution neural networks (CNN), while it still suffers interference from complex background and large-scale variation. To relieve the above challenges, this paper proposes a novel Multi-branch Progressive Embedding Network (MPENet) for crowd counting. Specifically, the proposed network mainly consists of two modules named Background Area Filter (BAF) and Sequential Multi-scale Modules (SMM), which are embedded with each other to generate higher-quality density maps. Firstly, the BAF model base on attention mechanism is proposed to distinguish crowd from background, which effectively avoids the model outputting positive predictions in the background region. Meanwhile, a multi-level supervision mechanism is proposed to generate more accurate attention maps. Besides, the SMM module is designed to be progressively embedded with scale context information so that the scale feature will be smooth and continuous. Finally, a novel multi-scale consistency structural loss is proposed to avoid pixel-level isolation due to Euclidean loss. The proposed method significantly improves counting accuracy, achieving Mean Absolute Error (MAE) of 57.6 and 6.9 on ShanghaiTechA and ShanghaiTechB respectively.

Enhancing personalized exercise recommendation with student and exercise portraits

The exercise recommendation system is emerging as a promising application in online learning scenarios, providing personalized recommendations to assist students with explicit learning directions. Existing solutions generally follow a collaborative filtering paradigm, while the implicit connections between students (exercises) have been largely ignored. In this study, we aim to propose an exercise recommendation paradigm that can reveal the latent connections between student-student (exercise-exercise). Specifically, a new framework was proposed, namely personalized exercise recommendation with student and exercise portraits (PERP). It consists of three sequential and interdependent modules: Collaborative student exercise graph (CSEG) construction, joint random walk, and recommendation list optimization. Technically, CSEG is created as a unified heterogeneous graph with students’ response behaviors and student (exercise) relationships. Then, a joint random walk to take full advantage of the spectral properties of nearly uncoupled Markov chains is performed on CSEG, which allows for full exploration of both similar exercises that students have finished and connections between students (exercises) with similar portraits. Finally, we propose to optimize the recommendation list to obtain different exercise suggestions. After analyses of two public datasets, the results demonstrated that PERP can satisfy novelty, accuracy, and diversity.

Programmable Optoelectronic Synaptic Transistors Based on MoS2/Ta2NiS5 Heterojunction for Energy-Efficient Neuromorphic Optical Operation.

The optoelectronic synaptic transistors with various functions, broad spectral perception, and low power consumption are an urgent need for the development of advanced optical neural network systems. However, it remains a great challenge to realize the functional diversification of the systems on a single device. 2D van der Waals (vdW) materials can combine unique properties by stacking with each other to form heterojunctions, which may provide a strategy for solving this problem. Herein, an all-2D vdW heterojunction-based programmable optoelectronic synaptic transistor based on MoS2/Ta2NiS5 heterojunctions is demonstrated. The device implements reconfigurable, multilevel non-volatile memory (NVM) states through sequential modulation of multiple optical and electrical stimuli to achieve broadband (532-808nm), energy-efficient (17.2 fJ), hetero-synaptic functionality in a bionic manner. The intrinsic working mechanisms of the photogating effect caused by band alignment and the interfacial trapping defect modulation induced by gate voltage are revealed by Kelvin-probe force microscopy (KPFM) measurements and carrier transport analysis. Overall, the (opto)electronic synaptic weight controllability for combined in-sensor and in-memory logic processors is realized by the heterojunction properties. The proposed findings facilitate the technical realization of generic all 2D hetero-synapses for future artificial vision systems, opto-logical systems, and Internet of Things (IoT) entities.

An empirical study of excitation and aggregation design adaptions in CLIP4Clip for video–text retrieval

CLIP4Clip model transferred from the CLIP has been the de-factor standard to solve the video clip retrieval task from frame-level input, triggering the surge of CLIP4Clip-based models in the video–text retrieval domain. In this work, we rethink the inherent limitation of widely-used mean pooling operation in the frame features aggregation and investigate the adaptions of excitation and aggregation design for discriminative video representation generation. We present a novel excitation-and-aggregation design, including (1) The excitation module is available for capturing non-mutually-exclusive relationships among frame features and achieving frame-wise features recalibration, and (2) The aggregation module is applied to learn exclusiveness used for frame representations aggregation. Similarly, we employ the cascade of sequential module and aggregation design to generate discriminative video representation in the sequential type. Besides, we adopt the excitation design in the tight type to obtain representative frame features for multi-modal interaction. The proposed modules are evaluated on three benchmark datasets of MSR-VTT, ActivityNet and DiDeMo, achieving MSR-VTT (43.9 R@1), ActivityNet (44.1 R@1) and DiDeMo (31.0 R@1). They outperform the CLIP4Clip results by +1.2% (+0.5%), +4.5% (+1.9%) and +9.5% (+2.7%) relative (absolute) improvements, demonstrating the superiority of our proposed excitation and aggregation designs. We hope our work will serve as an alternative for frame representations aggregation and facilitate future research.

Task-order control in dual-tasks: Only marginal interactions between conflict at lower levels and higher processes of task organization.

When simultaneously performing two tasks that share response properties, interference can occur. Besides general performance decrements, performance in the first task is worse when the second task requires a spatially incompatible response, known as the backward crosstalk effect (BCE). The size of this BCE, similar to congruency effects in conflict tasks, is subject to a sequential modulation, with a smaller BCE after incompatible compared to compatible trials. In the present study, we focus on a potential bidirectional interaction between crosstalk (and its resolution) at a lower level of task performance and higher-order processes of task organization. Two questions were of particular interest: First, do participants switch task order more frequently after a conflict-prone incompatible trial than after a compatible trial? Second, does changing task order influence the efficiency of conflict resolution, as indexed by the size of the sequential modulation of the BCE. Across four experiments, we only found marginal evidence for an influence of lower-level conflict on higher-order processes of task organization, with only one experiment revealing a tendency to repeat task order following conflict. Our results thus suggest practical independence between conflict and task-order control. When separating processes of task selection and task performance, the sequential modulation was generally diminished, suggesting that conflict resolution in dual-tasks can be disrupted by a deliberate decision about task order, or, alternatively, by a longer inter-trial interval. Finally, the study found a strong bias towards repeating the same task order across trials, suggesting that task-order sets not only impact task performance but also guide task selection.

Wavy-attention network for real-time cyber-attack detection in a small modular pressurized water reactor digital control system

Global interest in advanced reactors has been reignited by recent investments in small modular reactors and micro-reactor design. The use of digital devices is essential for meeting the size and modularity requirements of small modular reactor controls. By fully digitizing the small modular reactor control systems, critical information can be obtained to optimize control, reduce costs, and extend the reactor’s lifetime. However, the potential for cyber-attacks on digital devices leaves digital control systems vulnerable. To address this risk, this study presents a novel wavy-attention network for sensor attack detection in nuclear plants. The wavy-attention network comprises stacks of batch-normalized, dilated, one-dimensional convolution neural networks, and sequential self-attention modules, superior to conventional single-layer networks on sequence classification tasks. To evaluate the proposed wavy-attention network architecture, the International Atomic Energy Agency’s Asherah Nuclear Simulator and a false data injection toolbox found in the literature, both implemented in MATLAB/SIMULINK, are utilized. This approach leverages changes in process measurements to identify and classify cyber-attacks on priority signals using the proposed wavy-attention network. Three false data injection attacks are simulated on the simulator’s pressure, temperature, and level sensors to obtain representative process measurements. The wavy-attention network is trained and validated with normal and compromised process variables obtained from the simulator. The performance of the wavy-attention network to discriminate between the reactor states using the test set shows 99% accuracy, as opposed to other baseline models such as vanilla convolution neural networks, long short-term memory networks, and bi-directional long short-term memory networks with 90%, 77%, and 91% accuracy, respectively. An ablation study is also conducted to test the contribution of each component of the proposed architecture. The theoretical framework of the proposed wavy-attention network and its implementation for nuclear reactor digital sensor attack detection are discussed in this paper.

Electricity/light-heat-hygro multi-responsive soft luminescent systems for rewritable and programmable information display

Rewritable luminescent materials are attracting widespread attention for information storage and display as they can help to reduce environmental wastes caused by the abuse of single-use and transient systems. Nevertheless, since most reported systems are chemical-stimuli-responsive and single-stimuli-responsive, it remains challenging to achieve the rewritable and programmable display of multiple information in one single system, which is essential to enhance information richness and display intelligence. In this work, an ideal kind of multi-responsive soft luminescent display systems are reported, which consist of heat/humidity-responsive fluorescence-phosphorescence dual-emission film and stacked graphene assembly (SGA) film by using thin poly (dimethyl siloxane) (PDMS) as middle connecting layer. In such multilayer structure, the heat generated by electricity connection or light illumination of bottom SGA film can be instantly transferred across the middle PDMS layer to cause heat-triggered red-to-blue emission color change, or to evaporate water for blue-to-red emission color recovery. Owing to the “residual-free” nature of these electricity, light and water stimuli, multi-responsive rewritability is guaranteed to enable the reversible writing and complete erasure of various letter/pattern information. On this basis, robust electricity/light-heat-hygro multi-responsive soft luminescent display systems are demonstrated, in which diverse letter/pattern information could be facilely programmed to be successively displayed and then completely reset to enable another data writing/erasing cycle by sequential modulation of different environmental stimuli. This study brings programmable merit for rewritable luminescent display systems and is expected to inspire more intelligent ones.

Tumor Microenvironment Remodeling-Mediated Sequential Drug Delivery Potentiates Treatment Efficacy.

Toll-like receptor 7/8 agonists, such as imidazoquinolines (IMDQs), are promising for the de novo priming of antitumor immunity. However, their systemic administration is severely limited due to the off-target toxicity. Here, this work describes a sequential drug delivery strategy. The formulation is composed of two sequential modules: a tumor microenvironment remodeling nanocarrier (poly(l-glutamic acid)-graft-methoxy poly(ethylene glycol)/combretastatin A4, termed CA4-NPs) and an immunotherapy nanocarrier (apcitide peptide-decorated poly(l-glutamic acid)-graft-IMDQ-N3 conjugate, termed apcitide-PLG-IMDQ-N3 ). CA4-NPs, as a vascular disrupting agent, are utilized to remodel the tumor microenvironment for enhancing tumor coagulation and hypoxia. Subsequently, the apcitide-PLG-IMDQ-N3 could identify and target tumor coagulation through the binding of surface apcitide peptide to the GPIIb-IIIa on activated platelets. Afterward, IMDQ is activated selectively through the conversion of "-N3 " to "-NH2 " in the presence of hypoxia. The biodistribution results confirm their high tumor uptake of activated IMDQ (22.66%ID/g). By augmenting the priming and immunologic memory of tumor-specific CD8+ T cells, 4T1 and CT26 tumors with a size of ≈500 mm3 are eradicated without recurrence in mouse models.

Progressive deep snake for instance boundary extraction in medical images

Boundary extraction is meaningful in medical image analysis since it explicitly extracts the tissue/lesions boundary coordinates, which benefits the follow-up diagnosis processes. However, accurately extracting human tissue/lesions is challenging due to unclear target boundaries or pseudo boundaries, which could lead to inferior performance of existing boundary extraction and/or semantic segmentation methods. We propose a Progressive Deep Snake (PDS), which is designed with a multi-step contour evolution (MCE) and a sequential label generation module (SLGM) to explicitly extract the smooth and continuous boundary of different tissue/lesions even if they are unclear, with irregular shape, or occluded by pseudo boundaries. (1) MCE is designed to initialize and evolve the contours to the target boundaries, where a multi-step deep supervision strategy is designed to improve flexibility and accuracy. (2) SLGM is designed to generate easy-to-difficult supervision signals for the contour evolution, which provides targets of moderate difficulty in each step for better guiding the contour evolution. Experiments on a challenging verbal segment datasets show that our method achieves state-of-the-art performance (MR_AVBCE dataset: mIoU: 0.9234, mDice: 0.9598, mBoundF: 0.9241, mWcov: 0.9704). Our code is available at https://github.com/sysu19351118/PDS-pytorch.

Improving Continuous Sign Language Recognition with Consistency Constraints and Signer Removal

Deep-learning-based continuous sign language recognition (CSLR) models typically consist of a visual module, a sequential module, and an alignment module. However, the effectiveness of training such CSLR backbones is hindered by limited training samples, rendering the use of a single connectionist temporal classification loss insufficient. To address this limitation, we propose three auxiliary tasks to enhance CSLR backbones. First, we enhance the visual module, which is particularly sensitive to the challenges posed by limited training samples, from the perspective of consistency. Specifically, since sign languages primarily rely on signers’ facial expressions and hand movements to convey information, we develop a keypoint-guided spatial attention module that directs the visual module to focus on informative regions, thereby ensuring spatial attention consistency. Furthermore, recognizing that the output features of both the visual and sequential modules represent the same sentence, we leverage this prior knowledge to better exploit the power of the backbone. We impose a sentence embedding consistency constraint between the visual and sequential modules, enhancing the representation power of both features. The resulting CSLR model, referred to as consistency-enhanced CSLR, demonstrates superior performance on signer-dependent datasets, where all signers appear during both training and testing. To enhance its robustness for the signer-independent setting, we propose a signer removal module based on feature disentanglement, effectively eliminating signer-specific information from the backbone. To validate the effectiveness of the proposed auxiliary tasks, we conduct extensive ablation studies. Notably, utilizing a transformer-based backbone, our model achieves state-of-the-art or competitive performance on five benchmarks, including PHOENIX-2014, PHOENIX-2014-T, PHOENIX-2014-SI, CSL, and CSL-Daily. Code and models are available at https://github.com/2000ZRL/LCSA_C2SLR_SRM.

ResDNet: Efficient Dense Multi-Scale Representations With Residual Learning for High-Level Vision Tasks.

Deep feature fusion plays a significant role in the strong learning ability of convolutional neural networks (CNNs) for computer vision tasks. Recently, works continually demonstrate the advantages of efficient aggregation strategy and some of them refer to multiscale representations. In this article, we describe a novel network architecture for high-level computer vision tasks where densely connected feature fusion provides multiscale representations for the residual network. We term our method the ResDNet which is a simple and efficient backbone made up of sequential ResDNet modules containing the variants of dense blocks named sliding dense blocks (SDBs). Compared with DenseNet, ResDNet enhances the feature fusion and reduces the redundancy by shallower densely connected architectures. Experimental results on three classification benchmarks including CIFAR-10, CIFAR-100, and ImageNet demonstrate the effectiveness of ResDNet. ResDNet always outperforms DenseNet using much less computation on CIFAR-100. On ImageNet, ResDNet-B-129 achieves 1.94% and 0.89% top-1 accuracy improvement over ResNet-50 and DenseNet-201 with similar complexity. Besides, ResDNet with more than 1000 layers achieves remarkable accuracy on CIFAR compared with other state-of-the-art results. Based on MMdetection implementation of RetinaNet, ResDNet-B-129 improves mAP from 36.3 to 39.5 compared with ResNet-50 on COCO dataset.

FSASA: Sequential recommendation based on fusing session-aware models and self-attention networks

The recommendation system can alleviate the problem of ?information overload?, tap the potential value of data, push personalized information to users in need, and improve information utilization. Sequence recommendation has become a hot research direction because of its practicality and high precision. Deep Neural Networks (DNN) have the natural advantage of capturing comprehensive relations among different entities, thus almost occupying a dominant position in sequence recommendation in the past few years. However, as Deep Learning (DL)-based methods are widely used to model local preferences under user behavior sequences, the global preference modeling of users is often underestimated, and usually, only some simple and crude user latent representations are introduced. Therefore, this paper proposes a sequential recommendation based on Fusing Session-Aware models and Self-Attention networks (FSASA). Specifically, we use the Self-Attentive Sequential Recommendation (SASRec) model as a global representation learning module to capture long-term preferences under user behavior sequences and further propose an improved session-aware sequential recommendation model as a local learning representation module from user model the user?s dynamic preferences in the historical behavior, and finally use the Gated Recurrent Unit (GRU) module to calculate their weights. Experiments on three widely used recommendation datasets show that FSASA outperforms state-of-the-art baselines on two commonly used metrics.

A Single-Stage Dual-Output Regulating Rectifier With Sequential Pulse Frequency Modulation Control for Wireless Biomedical Applications

A Single-Stage Dual-Output Regulating Rectifier With Sequential Pulse Frequency Modulation Control for Wireless Biomedical Applications