Basic Modules Research Articles

MPE-HRNetL: A Lightweight High-Resolution Network for Multispecies Animal Pose Estimation

Animal pose estimation is crucial for animal health assessment, species protection, and behavior analysis. It is an inevitable and unstoppable trend to apply deep learning to animal pose estimation. In many practical application scenarios, pose estimation models must be deployed on edge devices with limited resource. Therefore, it is essential to strike a balance between model complexity and accuracy. To address this issue, we propose a lightweight network model, i.e., MPE-HRNet.L, by improving Lite-HRNet. The improvements are threefold. Firstly, we improve Spatial Pyramid Pooling-Fast and apply it and the improved version to different branches. Secondly, we construct a feature extraction module based on a mixed pooling module and a dual spatial and channel attention mechanism, and take the feature extraction module as the basic module of MPE-HRNet.L. Thirdly, we introduce a feature enhancement stage to enhance important features. The experimental results on the AP-10K dataset and the Animal Pose dataset verify the effectiveness and efficiency of MPE-HRNet.L.

Fault3DNnet: A lightweight 3D seismic fault detection network with bidirectional decoding

Fault detection from seismic data is a critical component of hydrocarbon exploration and production. In recent years, deep-learning techniques have made tremendous achievements and breakthroughs in seismic fault detection. Especially U-nets, represented by FaultSeg3D and its variants, are a dominant method for 3D deep-learning-based seismic fault interpretation. The incorporation of state-of-the-art network modules or training strategies with U-nets continuously improves the performance of fault detection for field seismic data, usually at the cost of increased computational intensity and hardware consumption. We design a lightweight bidirectional decoding network, Fault3DNnet, for fault detection from 3D seismic data volume. To make it simple and concise, our Fault3DNnet is equipped with basic conventional convolution and regular operation modules. The two decoding branches in Fault3DNnet could provide complementary information and improve fault detection performance. The superiority of Fault3DNnet is verified using publicly available synthetic and four field seismic data sets. The experimental results indicate that the method predicts fault structures with improved completeness and continuity compared with FaultSeg3D. Meanwhile, the parameter and computational quantity of the network are only 1/7 and 1/5 of FaultSeg3D, respectively. With the same graphics processing unit memory consumption, Fault3DNnet can handle the size of approximately six times the volume of 3D seismic data during inferencing as that of FaultSeg3D. It helps to improve prediction stability and decrease discontinuous artifacts at the stitching boundaries when predicting the large field data in chunks. In general, our Fault3DNnet delivers superior fault detection results while greatly reducing computational intensity and memory usage.

Construction of a prototype of the basic module of a scalable robotic system for simulation of composite surfaces for dynamic tests

The authors of the article developed a mathematical model of Stewart platform management. The angles of rotation of the servo shafts were obtained when setting various laws of motion of the mobile platform, in matrix mode. The hardware part of the prototype of the basic module of a scalable robotics system for modeling composite surfaces for dynamic tests was developed. The results of calculations of the maximum permissible load on the shafts of servos are presented. The CubeIDE settings for programming the STM32 microcontroller are given. The results of software development for the control of this prototype were obtained, namely, a library for controlling 6 servo drives under driver control was written. The problems that may arise during the above-described developments by future developers are considered, solutions to these problems are obtained. Detailed design features of this prototype are described, the control code of the servo drive is given; steps for further development of the final robotic system are considered

Non-intrusive identification of building loads using EDCA-ShuffleNetV2 with fused feature visualization

Abstract Non-intrusive load monitoring (NILM) enables monitoring of appliance operation status and energy consumption without additional meters, providing innovative data support for energy management. However, current NILM systems still face challenges such as insufficient load feature information and low load identification accuracy. To utilize the load feature information and improve the accuracy of load identification more effectively, this paper first selects loads with high energy consumption and high usage frequency as the research object, makes full use of the load feature information through feature fusion, and converts the fused typical load data into an image through the Gramian summation angular field, and further carries out the identification of loads that still have similarity after feature fusion. Secondly, this paper improves the ShuffleNetV2 model structure by embedding an efficient dual-channel attention (EDCA) module in the basic feature extraction module of ShuffleNetV2 to enhance it to extract adequate load feature information. A residual structure is also introduced to mitigate the information loss and gradient disappearance issues of the model. Finally, the superiority and effectiveness of the fused load features and the proposed model in load identification and typical scenario applications are validated using the iAWE and AMPDS datasets.

Neural architecture search for multi-sensor information fusion-based intelligent fault diagnosis

The multi-sensor information fusion-based intelligent fault diagnosis (MIF-IFD) method using deep learning aims to provide more accurate and reliable ways of identifying equipment health status than single-sensor methods. However, designing feature extraction modules in deep networks and determining information fusion execution layers still requires expertise and extensive experimentation, imposing high development costs. This limitation impedes the development and application of MIF-IFD. This paper proposes an architecture to mitigate the development challenges of MIF-IFD. Firstly, the basic module of the network consists of searchable feature extraction cells and a mask self-attention mechanism to achieve the extraction and fusion of multi-sensor information. Secondly, the optimal feature extraction cells and fusion starting layers are searched by constructing a continuous search space to compose the optimal fusion diagnostic structure, respectively. Finally, validation and analysis are conducted using two experimental datasets. The results demonstrate that the proposed method not only ensures diagnostic accuracy but also significantly improves development efficiency.

Redesign of a virtual reality basic life support module for medical training – a feasibility study

BackgroundHealthcare providers, including medical students, should maintain their basic life support (BLS) skills and be able to perform BLS in case of cardiac arrest. Research shows that the use of virtual reality (VR) has advantages such as improved accessibility, practice with lifelike situations, and real-time feedback during individual training sessions. A VR BLS module incorporating these advantages, called Virtual Life Support, has been developed especially for the medical domain. Virtual Life Support was collaboratively developed by software developers and stakeholders within the field of medical education. For this study, we explored whether the first version of this module capitalised on the advantages of VR and aimed to develop an understanding of barriers to feasibility of use.MethodsThis study was conducted to assess the feasibility of employing Virtual Life Support for medical training and pinpoint potential obstacles. Four groups of stakeholders were included through purposive sampling: physicians, BLS instructors, educational experts, and medical students. Participants performed BLS on a BLS mannequin while using Virtual Life Support and were interviewed directly afterwards using semi-structured questions. The data was coded and analysed using thematic analysis.ResultsThematic saturation was reached after seventeen interviews were conducted. The codes were categorised into four themes: introduction, content, applicability, and acceptability/tolerability. Sixteen barriers for the use of Virtual Life Support were found and subsequently categorised into must-have (restraining function, i.e. necessary to address) and nice to have features (non-essential elements to consider addressing).ConclusionThe study offers valuable insights into redesigning Virtual Life Support for Basic Life Support training, specifically tailored for medical students and healthcare providers, using a primarily qualitative approach. The findings suggest that the benefits of virtual reality, such as enhanced realism and immersive learning, can be effectively integrated into a single training module. Further development and validation of VR BLS modules, such as the one evaluated in this study, have the potential to revolutionise BLS training. This could significantly improve both the quality of skills and the accessibility of training, ultimately enhancing preparedness for real-life emergency scenarios.

Research on the Modular Design Method and Application of Prefabricated Residential Buildings

As one of the key ways to realize the industrialization and green development of construction, prefabricated construction is conducive to saving resources and energy and improving labor productivity and quality. Aiming to solve the problem of the lack of standardization in the design of prefabricated residential buildings, which leads to the components not being universally used and the industrial characteristics not being fully embodied, while excessive standardization leads to a lack of personalization and flexibility, the modular design theory is applied to the standardized design of prefabricated residential buildings in this study. The application route of modular design theory in the standardized design is constructed, that is, “system decomposition—module design—module combination”. Taking residential buildings within a height of 54 m as an example, each basic functional module is standardized and combined into standard plans. At the same time, the functional space module design based on modular coordination and the module combination design based on the trinity of “modulus, pattern, and mode” are discussed. This research is of great significance for giving full play to the comprehensive benefits of prefabricated concrete structures in quality improvement, cost reduction, and rapid assembly.

From Dependent to Self-Directed Learning in Medical Education: Can Online Modular Intervention Facilitate the Transition?

Background: Indian medical graduates are expected to be lifelong learners, and medical teachers need to find newer and more engaging ways to help them hone their lifelong learning skills. Online learning can be a stepping stone toward this goal. Methods: We conducted a pre-post design interventional study in which 22 phase two MBBS students rated their self-directedness in learning using the Self-Rating Scale of Self-directed Learning (SRSSDL) at two points in time—before and after completing a 10-week online basic psychopharmacology module. Participant feedback was obtained regarding various aspects of the online modular learning experience. Results: The participants’ mean SRSSDL scores before and after the intervention were 231.31 ± 26.64 and 227.31 ± 31.33. The difference is not significant. The students’ scores at both points in time placed them in the “high” range of level of self-directed learning, leaving less realistic scope for a large change in the given time frame. Participant feedback was encouraging, with 19 out of 22 reporting that the overall learning experience was a “positive” one. Conclusions: Second-year medical students’ self-reported self-directedness is high. Online modular learning may be a feasible addition to medical education, even though it may not increase self-directedness in the short term. A limited sample size can lead to type 2 errors.

CvFormer: Cross-view transFormers with pre-training for fMRI analysis of human brain

In recent years, functional magnetic resonance imaging (fMRI) has been widely utilized to diagnose neurological disease, by exploiting the region of interest (RoI) nodes as well as their connectivities in human brain. However, most of existing works only rely on either RoIs or connectivities, neglecting the potential for complementary information between them. To address this issue, we study how to discover the rich cross-view information in fMRI data of human brain. This paper presents a novel method for cross-view analysis of fMRI data of the human brain, called Cross-view transFormers (CvFormer). CvFormer employs RoI and connectivity encoder modules to generate two separate views of the human brain, represented as RoI and sub-connectivity tokens. Then, basic transformer modules can be used to process the RoI and sub-connectivity tokens, and cross-view modules integrate the complement information across two views. Furthermore, CvFormer uses a global token for each branch as a query to exchange information with other branches in cross-view modules, which only requires linear time for both computational and memory complexity instead of quadratic time. To enhance the robustness of the proposed CvFormer, we propose a two-stage strategy to train its parameters. To be specific, RoI and connectivity views can be firstly utilized as self-supervised information to pre-train the CvFormer by combining it with contrastive learning and then fused to finetune the CvFormer using label information. Experiment results on two public ABIDE and ADNI datasets can show clear improvements by the proposed CvFormer, which can validate its effectiveness and superiority.

SRARDA: A lightweight adaptive residual dense attention generative adversarial network for image super-resolution

Image super-resolution (SR) is the task of inferring a high resolution (HR) image from one/multiple single low resolution (LR) input(s). Traditional networks are evaluated by pixel-level metrics such as Peak-Signal-to-Noise Ratio (PSNR) etc., which do not always align with human perception of image quality. They often produce excessively smooth images that lack high-frequency texture and appear unnatural. Therefore, in this paper, we propose a lightweight adaptive residual dense attention generative adversarial network (SRARDA) for image SR. Firstly, our generator adopts the residual in residual (RIR) structure but redesigns the basic module. By using dynamic residual connection (ARC) to dynamically adjust the importance of residual and main paths, we design a novel adaptive residual dense attention block (ARDAB) that enhances the feature extraction capability of the generator. In addition, we build a high-frequency filtering unit (HFU) to extract more high-frequency features from the LR space. Finally, to fully utilize the discriminator, we use WGAN to compute the difference between the HR image and the reconstructed image. Experiments demonstrate that SRARDA effectively addresses the issue of excessive smoothing in reconstructed images, while also enhancing visual quality.

Fuzzy inference system enabled neural network feedforward compensation for position leap control of DC servo motor

To improve dynamic performance and steady-state accuracy of position leap control of the direct current (DC) servo motor, a fuzzy inference system (FIS) enabled artificial neural network (ANN) feedforward compensation control method is proposed in this study. In the method, a proportional-integral-derivative (PID) controller is used to generate the baseline control law. Then, an ANN identifier is constructed to online learn the reverse model of the DC servo motor system. Meanwhile, the learned parameters are passed in real-time to an ANN compensator to provide feedforward compensation control law accurately. Next, according to system tracking error and network modeling error, an FIS decider consisting of an FI basic module and an FI finetuning module is developed to adjust the compensation quantity and prevent uncertain disturbance from undertrained ANN adaptively. Finally, the feasibility and efficiency of the proposed method are verified by the tracking experiments of step and square signals on the DC servo motor testbed. Experimental results show that the proposed FIS-enabled ANN feedforward compensation control method achieves lower overshoot, faster adjustment, and higher precision than other comparative control methods.

Evaluating the Effectiveness of Assessment Methods in Moulay Ismail University: Insights from Postgraduate Students

The present study evaluates and measures the effectiveness of the assessment practices promoted in the LMD system (Licence, Master and Doctorate) implemented in Moroccan universities from the perspective of postgraduate students at the open-access institutions affiliated with Moulay Ismail University (UMI), namely the School of Arts and Humanities (FLSH), the School of Sciences (FS), the School of Law, Economics and Social Sciences (FSJES), and the Polydisciplinary School of Errachidia (FP). To achieve this objective, the study deployed a mixed-methods; two-phase case study design. In the initial quantitative phase, a questionnaire was administered to 742 postgraduate students to investigate their evaluation of the assessment practices promoted in the LMD system. Further, in the follow-up qualitative phase, interviews with 28 doctoral students were conducted to extend and explain the quantitative results. It has been found that continuous assessment is not used by most teachers and that the final exams are the only method used by professors to assess students' performance. Likewise, the study revealed that the compensation system negatively affects the quality of higher education training by allowing students to pass and study more difficult modules while they have not learned the necessary prerequisites from the basic modules. Equally significant, the study found that most students struggle with the end-of-study research module, blaming their supervisors who do not provide sufficient or appropriate supervision. In conclusion, this paper offers several pedagogical implications for stakeholders and teachers at the UMI to make assessment more effective.

Non‐Hermitian Skin Effect in Non‐Hermitian Optical Systems

AbstractThe development of non‐Hermitian (NH) physics in nonconservative systems has led to various interesting wave phenomena, including the emergence of the non‐Hermitian skin effect (NHSE). NHSE is a topological effect caused by the nontrivial winding number of the energy spectrum, which results in many bulk eigenmodes localized on low‐dimensional boundaries under open boundary conditions. Although this phenomenon is striking, the flexible and dynamic tuning of NHSE has not been reviewed in optics systems. As an ideal NH platform, an optical system can introduce artificially constructed gains and losses and flexibly design optical structures to control the transmission and regulation of light. This exploration based on NH optical systems deepens the understanding of NHSE and facilitates technological breakthroughs in high‐energy directional localized optics applications. In this review, the focus is on the latest progress and applications of NHSE in NH optical systems, including the basic theory and local property modulation methods. Finally, possible future research directions for NHSE based on NH optical systems are discussed.

PandaMom – Feasibility and acceptability of an internet- and mobile-based intervention to enhance peripartum mental well-being and to prevent postpartum depression

BackgroundMental disorders during pregnancy and the postpartum period can have far-reaching consequences. To enhance peripartum mental well-being and prevent peripartum mental disorders, internet- and mobile-based interventions appear promising. They can overcome help-seeking barriers associated with face-to-face conditions and have proven to be effective. However, previous findings are scarce and mixed. The primary objectives of this study were to assess the feasibility and acceptability of an internet-based program aimed at enhancing peripartum mental well-being and preventing postpartum depression. MethodsIn total, 149 pregnant, German-speaking women were assigned to the internet-based intervention PandaMom. The program comprises a total of 10 basic and supplementary modules related to pregnancy and postpartum, based on cognitive-behavioral principles. Additionally, PandaMom offers professional, individualized guidance and a moderated group-chat. Assessments were conducted at baseline (pre-intervention), as well as two and five weeks postpartum. The primary outcomes included feasibility, user satisfaction, and adherence to the intervention. Secondary outcomes included depressive symptomatology, anxiety and stress. ResultsPandaMom was found to be feasible, and evaluation of module content and length satisfaction indicated that the intervention was well accepted. Nearly half of the participants utilized the guidance service by responding to individual messages from their intervention moderator. Regarding working alliance, participants reported a strong bond with their intervention moderator. Of the 149 participants, 132 logged into the platform at least once. 113 participants accessed at least one module, with an average of 4.7 modules opened per participant. However, only 16 participants completed the basic modules. ConclusionThe findings of this study support previous evidence that internet-and mobile-based interventions are feasible and acceptable during pregnancy and the postpartum period. Further research is needed to address the challenge of low adherence and to evaluate the efficacy of PandaMom.

Deep learning model for the deformation prediction of concrete dams under multistep and multifeature inputs based on an improved autoformer

The long-term prediction of deformation in concrete dams is a critical requirement for maintaining their structural integrity over time in practical management scenarios. While most existing models predominantly address short-term and medium-term predictions, there remains a significant challenge in establishing reliable correlations within complex long-term temporal patterns and varying environmental factors. Consequently, there is limited research on multistep prediction. To solve these problems, an Autoformer-based model for multistep and multifeature dam deformation prediction is proposed. In the proposed model, the sequence decomposition of dam monitoring data is used as the basic construction module inside the depth model. Then, Deep Automatic Correlation (Deep-AutoCorrelation) mechanism is employed to obtain the long-term dependence between dam deformation and environmental load. The encoder-decoder structure integrates time series decomposition and the Deep-AutoCorrelation to predict the multistep deformation of the dam. In the multistep deformation prediction of the dam, the improved Autoformer model demonstrates state-of-the-art performance, resulting in a 48% increase in prediction accuracy across five monitoring point datasets when compared to six benchmark models. Experimental results indicate that this method effectively addresses the challenge of predicting long-term deformation of dams subjected to complex environmental loads. It demonstrates robust capabilities in extracting long-term temporal features and avoids the increase of computational complexity caused by deeper time extraction.

Results of the implementation of the DP-TRANSFERS project in Catalonia: a translational method to improve diabetes screening and prevention in primary care

IntroductionDP-TRANSFERS is a translational lifestyle intervention project, which follows a previous protocol described in the DE-PLAN-CAT study. ObjectiveAnalyze the feasibility of reproducing the intensive intervention and estimating the effect of translation in real conditions of clinical practice in primary care. MethodologyImplementation of the face-to-face group intervention adjusted to 2 years. After screening, the intervention consisted of a basic module and a continuity module. Stratifying by clusters (health centers), a representative sample (centers, professionals and participants) was evaluated (FINDRISC > 11 and/or prediabetes) from 2016 to 2020. The effect of the intervention on the incidence of diabetes was analyzed. ResultsThe intervention, feasible in 95 of 123 centers, involved 343 of 647 professionals. Of 2381 subjects screened, 1713 participated in the basic module, with 1186 participants completing the first year and 776 completing the second. 121 participants (7.06%) were diagnosed with diabetes: 77 (4.49%) during the first year; 44 (2.57%) during the second.The bivariate analysis showed that those subjects in whom diabetes affected differed in: previous glycemic status, A1c, HDL-cholesterol, FINDRISC score and adherence to the Mediterranean diet, and in the differences between the beginning and end of the study of: body weight, BMI and abdominal circumference. ConclusionsThe intensive intervention substantially reduced (23.6%) the incidence of diabetes compared to that previously estimated in standardized intervention. The following acted as protective factors: a better glycemic status, lower baseline risk, elevated HDL-cholesterol, or achieving a reduction in weight or abdominal circumference during the study.

Advances in artificial cells based on microfluidic chips

One of the main goals of synthetic biology is to build artificial cells in a bottom-up manner, which not only facilitates the deep understanding of the origin of life and cell function but also plays a critical role in the research fields such as the development of artificial cell chassis, tissue models, engineering drug delivery systems, and drug screening tools. However, achieving this goal is extremely challenging. The complexity of cell structures and the miniaturization and diversity of basic modules pose high requirements for the construction methods. The microfluidic chip, as an advanced microanalysis system, serves as an effective tool for building artificial cells. It can accurately control the structure and local microenvironment of artificial cells, becoming the preferred approach for the current research on synthetic life. This article reviewed the methods of constructing, manipulating, and analyzed artificial cells based on microfluidic chips, emphasized the importance of the microenvironment for life systems and artificial self-sustaining systems. In addition, this article demonstrated the wide applications of artificial cells in multiple critical biomedical fields. Exploring the advantages, disadvantages, and application performance of different microfluidic methods can enrich our knowledge about artificial cell research. Finally, we made an outlook on the development of artificial cell research based on microfluidics, expecting that this field can achieve greater breakthroughs and progress.

Challenges and solutions in the mechanical design of the KM3NeT ARCA Base Module

The goal of KM3NeT ARCA is to search for astrophysical neutrinos with a neutrino telescope operating on the bottom of the Mediterranean Sea, at 3500 m depth, for more than 15 years. A key component of the neutrino telescope is the Base Module (BM) which is built with a titanium cylindrical container hosting optics, electronics and power boards. From a mechanical point of view the BM must be water tight and resistant to an external water pressure up to 350 bar, be able to stand all mechanical stresses along its life, provide excellent corrosion resistance for 15 years at least in sea water and be able to host all internal components safely, by guaranteeing an adequate heat dissipation. The tests performed for the qualification of the mechanical design are presented here. The first BM is successfully operating underwater since 2015.

An Interpretable Light Attention–Convolution–Gate Recurrent Unit Architecture for the Highly Accurate Modeling of Actual Chemical Dynamic Processes

To equip data-driven dynamic chemical process models with strong interpretability, we develop a light attention–convolution–gate recurrent unit (LACG) architecture with three sub-modules—a basic module, a brand-new light attention module, and a residue module—that are specially designed to learn the general dynamic behavior, transient disturbances, and other input factors of chemical processes, respectively. Combined with a hyperparameter optimization framework, Optuna, the effectiveness of the proposed LACG is tested by distributed control system data-driven modeling experiments on the discharge flowrate of an actual deethanization process. The LACG model provides significant advantages in prediction accuracy and model generalization compared with other models, including the feedforward neural network, convolution neural network, long short-term memory (LSTM), and attention-LSTM. Moreover, compared with the simulation results of a deethanization model built using Aspen Plus Dynamics V12.1, the LACG parameters are demonstrated to be interpretable, and more details on the variable interactions can be observed from the model parameters in comparison with the traditional interpretable model attention-LSTM. This contribution enriches interpretable machine learning knowledge and provides a reliable method with high accuracy for actual chemical process modeling, paving a route to intelligent manufacturing.

Topological traits are associated with the prevalence of IGP modules in empirical food webs

Exploring how food webs are assembled from basic modules is charming and crucial for understanding how communities are self-organized. As one of the basic modules, intraguild predation (IGP) consists of a prey being consumed by both an intermediate and a top predator, with the former also being consumed by the latter (thus encompassing both predation and competition). This interaction has been shown to govern food web stability, and therefore underpin the organization of network structures. While some studies have been made in understanding the factors and mechanisms behind the prevalence of IGP modules in food webs, the specific role of food web topological structures in relation to these modules remains largely unexplored and is not well understood. Here, 103 food webs were analyzed, and we found that the number of modules in each food web was largely determined by taxon richness and connectance. After controlling richness and connectance, the specific scale-free pattern and core-periphery structure of empirical food webs explains the higher prevalence of IGP modules in empirical food webs better than by chance. Lastly, the loss of taxa which supported large number of IGP modules would lead to serious damage to food web robustness, indicating the keystone role of these taxa in maintaining food web structure and stability. Our results provide new insight into the assembly of empirical food webs from the perspective of IGP modules.