Parallel Modules Research Articles

Dual-decoding branch contrastive augmentation for image manipulation localization

With the rapid development of image editing techniques, forensic analysis to detect malicious image manipulation has become an important research topic. The current image manipulation detection and localization methods can accommodate diverse forms of tampering. However, their approach to handling various tampering detection types is limited to a uniform regression pathway. This approach fails to recognize the unique characteristics of copy-move tampering, which is significantly different from other tampering types. Employing a generic detection methodology indiscriminately poses the risk of confusing the training regression trajectory of the deep learning models. To mitigate this challenge, this paper introduces a novel framework featuring a dual-decoding branch structure specifically designed to augment features pertinent to copy-move tampering types. Moreover, it facilitates the detection of tampered regions, irrespective of the tampering type, within the main branch. To achieve this goal, we first introduce a contrastive augmentation module in the encoder, which maximizes the feature space distance between the manipulation regions and pristine regions. Next, we design a parallel attention module to extract more diverse multiscale features. Moreover, we introduce a constrained shifted-window dual attention module to extract tampering noise features. In the decoder, we design a dual-decoding branch to capture both the homologous and tampering features, and we employ contrastive learning to minimize the feature space distance of the homologous regions for copy-move manipulation detection. Finally, we design a category normalization loss function to balance the model’s attention across each category. Extensive experiments demonstrate that the proposed approach achieves state-of-the-art performance on various benchmark datasets.

PDCA-Net: Parallel dual-channel attention network for polyp segmentation

Accurate segmentation of polyps in colonoscopy images is crucial for the diagnosis and cure of colorectal cancer. Although various deep learning methods have been proposed and have shown promising performance, accurately distinguishing between polyp and mucosal boundaries remains a challenge. In this work, we propose a Parallel Dual-Channel Attention Network (PDCA-Net) for polyp segmentation. This method utilizes the mapping transformations to adaptively encapsulate the global dependency from superpixel into pixels, enhancing the model’s ability to localize foreground and background regions. Specifically, we first design a parallel spatial and channel attention fusion module to capture the global dependencies at the superpixel level from the spatial and channel dimensions. Furthermore, an adaptive associative mapping module is proposed to encapsulate the global dependencies of superpixels into each pixel through a coarse-to-fine learning strategy. Extensive experiments demonstrate that the proposed PDCA-Net effectively improves the segmentation performance and achieves new state-of-the-art results (i.e., 0.815, 0.936, 0.945, and 0.838 mDice, 0.744, 0.891, 0.900, and 0.765 mIoU on the ETIS, Kvasir-SEG, CVC-ClinicDB, and CVC-ColonDB). Our code is available at https://github.com/lzucg/PDCA-Net.

Mechanism and Control Strategies for Current Sharing in Multi-Chip Parallel Automotive Power Modules

Multi-chip parallel power modules are highly favored in applications requiring high capacity and high switching frequency. However, the dynamic current imbalance between parallel chips caused by asymmetric layouts limits the available capacity. This paper presents a method to optimize dynamic current distribution by adjusting the lengths and connection points of bond wires. For the first time, a response surface model and nonlinear constraint optimization algorithm are introduced, along with parameter analysis based on finite element methods, to establish the response surface models for the parasitic inductance of bond wires and DBC (direct bonded copper). By leveraging the optimization goals for parasitic inductance and the analytical expressions of all response surfaces, the dynamic current sharing issue was transformed into a nonlinear constrained optimization problem. The solution to this optimization problem identified the optimal connection points for the bond wires, enhancing dynamic current sharing performance. Simulations and experiments were conducted, revealing that the optimized automotive-grade module exhibited a significant reduction in current differences between parallel branches, from 41.7% to 5.03% compared with the original design. This indicated that the proposed optimization scheme for adjusting bond wire connection points could significantly mitigate current disparities, thereby markedly improving current distribution uniformity.

Detection of coronary heart disease based on heart sound and hybrid Vision Transformer

As a non-invasive examination method, phonocardiogram (PCG) has been widely used to assist in the diagnosis of cardiovascular diseases. Coronary heart disease (CHD) is the most common cardiovascular disease and has become one of the leading causes of death worldwide. Utilizing computer-aided PCG analysis for CHD detection can assist doctors in diagnosis, reducing the huge burden of disease. Inspired by the success of the deep convolutional neural network (CNN) and the Vision Transformer (ViT) in computer vision, this study proposes a hybrid Convolution-Transformer neural network (HCTNN) and verifies the feasibility of a Transformer-based method for CHD detection in a clinical PCG dataset. Firstly, clinical heart sound recordings are preprocessed and transformed into continuous wavelet transform (CWT) spectrograms to serve as inputs for the following model. In HCTNN, a pruned ViT module is designed to encode and aggregate global features, while a parallel CNN module is designed to extract local features. Subsequently, a feature fusion module based on a reweighting mechanism is employed to adaptively adjust the fusion of global and local features, and the final classification is performed by a multilayer perceptron. The experimental analysis proves that the proposed HCTNN achieves CHD detection with an accuracy of 94.24%, a sensitivity of 96.32%, a specificity of 91.82%, a precision of 93.19%, and an F1 score of 94.73%, outperforming standalone ViT and several advanced CNN models. This study demonstrates the effectiveness of methods based on PCG and Transformer models in detecting CHD, which has broad research prospects.

TF-BAPred: A Universal Bioactive Peptide Predictor Integrating Multiple Feature Representations

Bioactive peptides play essential roles in various biological processes and hold significant therapeutic potential. However, predicting the functions of these peptides is challenging due to their diversity and complexity. Here, we develop TF-BAPred, a framework for universal peptide prediction incorporating multiple feature representations. TF-BAPred feeds original peptide sequences into three parallel modules: a novel feature proposed in this study called FVG extracts the global features of each peptide sequence; an automatic feature recognition module based on a temporal convolutional network extracts the temporal features; and a module integrates multiple widely used features such as AAC, DPC, BPF, RSM, and CKSAAGP. In particular, FVG constructs a fixed-size vector graph to represent the global pattern by capturing the topological structure between amino acids. We evaluated the performance of TF-BAPred and other peptide predictors on different types of peptides, including anticancer peptides, antimicrobial peptides, and cell-penetrating peptides. The benchmarking tests demonstrate that TF-BAPred displays strong generalization and robustness in predicting various types of peptide sequences, highlighting its potential for applications in biomedical engineering.

Evaluation of the effect of disposal of landfill leachate in a sewage treatment plant composed of stabilization ponds

This study aimed to evaluate the combined treatment system of domestic sewage and leachate by serial stabilization ponds in a Wastewater Treatment Plant (WWTP). The WWTP had two parallel pond modules, one receiving (module II—pond M10) and the other not receiving leachate (module I—pond M5). Physical, chemical, and biological monitoring parameters collected from January 2017 to December 2021 were evaluated. High reducing efficiency for ammonium nitrogen was found, with an average of 95.7 ± 2.5%. BOD5 and COD can be considered statistically different for a 95% confidence level in the two pond modules. The phytoplankton community was composed of 46 taxa distributed in four taxonomic classes: 6 belonging to the Cyanobacteria group (13%), 22 to the Green Algae group (48%), 4 to the Diatoms group (9%), and 14 to the Phytoflagellate group (30%). No ecotoxicity was detected for the evaluated pond effluent samples. Thus, although the insertion of leachate impacted the quality of the final effluent of the ponds and the organic matter, it may not have affected the effluents’ ecotoxicity. This research helps to understand the impacts of long-term leachate insertion in a real treatment system using stabilization ponds for combined treatment. In addition to the usual efficiency evaluation parameters, ecotoxicological tests and phytoplankton are also evaluated.

Baseline-free assisted lamb wave-based damage detection in CFRP composites using graph convolutional networks and Transformer models

Carbon fiber-reinforced polymer (CFRP) composites are widely used in aerospace but are susceptible to damage, threatening structural safety. Thus, developing an effective structural health monitoring system is essential. This study presents CFRP-former, a novel deep learning framework that detects damage in CFRP composites by constructing spatio-temporal representation graphs of Lamb waves (LWs). The CFRP-former comprises three key modules: graph convolutional neural network (GCNN), a multi-layer perceptron (MLP), and a Transformer Encoder. The GCNN models the topological relationships between LW nodes and sensors. The MLP reduces data dimensionality while preserving key information for subsequent analysis. Meanwhile, the Transformer Encoder, employing multi-head attention mechanisms, captures global time-series patterns in the LW signals. Additionally, the CFRP-former framework includes a parallel module that separately predicts both the size and location of damage. Experiments using only four sensors show that CFRP-former achieves high accuracy in detecting and quantifying damage, with robustness confirmed through ablation tests.

Simultaneous Photonics Generation of Multiple Chirp Rate Microwave Waveform With Manifold Frequency Multiplying Capability by Using Triple Parallel Mach-Zehnder Modulators

Simultaneous Photonics Generation of Multiple Chirp Rate Microwave Waveform With Manifold Frequency Multiplying Capability by Using Triple Parallel Mach-Zehnder Modulators

Convolutional Transformer-Based Cross Subject Model for SSVEP-Based BCI Classification.

Steady-state visual evoked potential (SSVEP) is a commonly used brain-computer interface (BCI) paradigm. The performance of cross-subject SSVEP classification has a strong impact on SSVEP-BCI. This study designed a cross subject generalization SSVEP classification model based on an improved transformer structure that uses domain generalization (DG). The global receptive field of multi-head self-attention is used to learn the global generalized SSVEP temporal information across subjects. This is combined with a parallel local convolution module, designed to avoid oversmoothing the oscillation characteristics of temporal SSVEP data and better fit the feature. Moreover, to improve the cross-subject calibration-free SSVEP classification performance, an DG method named StableNet is combined with the proposed convolutional transformer structure to form the DG-Conformer method, which can eliminate spurious correlations between SSVEP discriminative information and background noise to improve cross-subject generalization. Experiments on two public datasets, Benchmark and BETA, demonstrated the outstanding performance of the proposed DG-Conformer compared with other calibration-free methods, FBCCA, tt-CCA, Compact-CNN, FB-tCNN, and SSVEPNet. Additionally, DG-Conformer outperforms the classic calibration-required algorithms eCCA, eTRCA and eSSCOR when calibration is used. An incomplete partial stimulus calibration scheme was also explored on the Benchmark dataset, and it was demonstrated to be a potential solution for further high-performance personalized SSVEP-BCI with quick calibration.

LPI Radar Waveform Recognition Based on Hierarchical Classification Approach and Maximum Likelihood Estimation.

The importance of information gathering is emphasized to minimize casualties and economic losses in warfare. Through electronic warfare, which utilizes electromagnetic waves, it is possible to discern the enemy's intentions and respond accordingly, thereby leading the battle advantageously. Consequently, related research is actively underway. The development of various radar signal modulation techniques has revealed limitations in the existing modulation recognition methods, necessitating the development of distinguishing features to overcome these limitations. This paper proposes and analyzes distinguishing features that can differentiate various modulation schemes. Eleven distinguishing features were employed, and twenty-two types of modulated signals, including analog, digital, and composite modulation, were classified using hierarchical classification approach and maximum likelihood estimation (MLE). The proposed method achieves a recognition performance of 99.76% at an SNR of 20 dB and 98.45% at an SNR of 8 dB.

A multi‐path residual network for image denoising based on edge prior information

AbstractEdge‐preserving denoising is important in image analysis. However, most existing methods suffer from smoothing and loss of high frequency detail at the edges. Aiming at this issue, a multi‐path residual network using edge prior information is proposed. Specifically, the edge is first recovered by a newly designed parallel edge extraction module. Then it is used as a priori to generate affine transform parameters to guide the weight distribution between noise and edge, so that the shallow feature extraction pays more attention to preserving edges. In the deep feature extraction stage, a multi‐scale attention unit is designed to provide the spatial neighbourhood information of the feature points to filter and activate the deep features, which further enhances the ability of the network to extract fine‐grained noisy features. Finally, by taking the difference between the noisy image and the noise feature obtained with residual learning, the denoised image is produced. Extensive experiments show that the PSNR is improved by 0.00–0.78 dB on grey image denoising and 0.00–2.69 dB on colour image denoising compared with others. The denoised image has a better ability to preserve image edges, high frequency details and visual perception.

Effect of module configurations on the performance of parallel-connected lithium-ion battery modules

To meet the power and energy of battery storage systems, lithium-ion batteries have to be connected in parallel to form various battery modules. However, different single module collector configurations (SCCs) and unavoidable interconnect resistances lead to inhomogeneous currents and state-of-charge (SoC) within the module, thereby significantly reducing the module lifetime and safety. Therefore, the correlation of SCCs to module inhomogeneous currents and SoC are worth investigating. Based on mathematical permutations and combinations calculations, all SCCs are firstly defined and obtained in this paper. Then, two parallel modules are developed in Matlab/Simscape based on fresh cell and degradation cell separately, which are used to quantify the influence of all SCCs. Furthermore, multiple module collector configurations (MCCs) and Urban Dynamometer Driving Schedule (UDDS) working condition are considered in the analysis. Results indicate that both module collectors connected to same edge cell of module cause the most inhomogeneity. The maximum SoC difference under the UDDS is only 46 % of the 0.5C. Moreover, MCCs can reduce inhomogeneous currents and SoC within the parallel module, while not all MCCs perform better homogeneous performance than SCCs. Finally, the effect of changes in SCCs on degradation module inhomogeneous currents and SoC is verified as same as for fresh cells parallel module. Based on the above results, the optimal single module collector configuration of the N cells parallel module is obtained through mathematical analysis, which can greatly improve the currents and SoC homogeneity of the module.

Image segmentation of tunnel water leakage defects in complex environments using an improved Unet model

Computer vision technology provides an intelligent means for detecting tunnel water leakage areas. However, the accuracy of defect feature extraction and segmentation is limited by factors such as insufficient lighting and environmental interference inside tunnels. To address the problem, this paper proposes a tunnel water leakage area segmentation network model called Customized Side Guided-Unet (CSG-Unet), using Unet as the baseline model. The main contributions are: (1) To improve the accuracy of water leakage area extraction, a customized side guided term is introduced to direct the net’s attention to the changes in light and shade within the image. A parallel attention network module is designed to extract internal information from the guided term. Subsequently, a strengthened channel attention module aggregates the guided term and the original information to achieve accurate segmentation of water leakage areas; (2) To address the scarcity of tunnel water leakage area datasets, a basic dataset is constructed by collecting data from open-source datasets and manually gathered data in tunnels. On this basis, perspective transformation is used to change the camera viewpoint, gaussian noise is randomly added to the images in the dataset to simulate images taken in dimly lit scenes, thereby expanding the dataset and enhancing the network’s generalization. The CSG-Unet network was trained using the constructed training set, achieving a mean Intersection over Union (mi IoU) of 85.54%, a mean Dice coefficient (mi Dice) of 85.26%, and a mean Pixel Accuracy (mi PA) of 90.85%. Compared to its baseline network, U-Net (tiny), these metrics show an improvement of over 3.2% in each indicator. Finally, a visual comparison between the improved network and the baseline network further confirms that the proposed model can effectively adapt to the segmentation of water leakage areas in complex environments.

Early diagnosis of Alzheimer’s disease using a group self-calibrated coordinate attention network based on multimodal MRI

Convolutional neural networks (CNNs) for extracting structural information from structural magnetic resonance imaging (sMRI), combined with functional magnetic resonance imaging (fMRI) and neuropsychological features, has emerged as a pivotal tool for early diagnosis of Alzheimer’s disease (AD). However, the fixed-size convolutional kernels in CNNs have limitations in capturing global features, reducing the effectiveness of AD diagnosis. We introduced a group self-calibrated coordinate attention network (GSCANet) designed for the precise diagnosis of AD using multimodal data, including encompassing Haralick texture features, functional connectivity, and neuropsychological scores. GSCANet utilizes a parallel group self-calibrated module to enhance original spatial features, expanding the field of view and embedding spatial data into channel information through a coordinate attention module, which ensures long-term contextual interaction. In a four-classification comparison (AD vs. early MCI (EMCI) vs. late MCI (LMCI) vs. normal control (NC)), GSCANet demonstrated an accuracy of 78.70%. For the three-classification comparison (AD vs. MCI vs. NC), it achieved an accuracy of 83.33%. Moreover, our method exhibited impressive accuracies in the AD vs. NC (92.81%) and EMCI vs. LMCI (84.67%) classifications. GSCANet improves classification performance at different stages of AD by employing group self-calibrated to expand features receptive field and integrating coordinated attention to facilitate significant interactions among channels and spaces. Providing insights into AD mechanisms and showcasing scalability for various disease predictions.

Dual-phase interface engineering via parallel modulation strategy for highly reversible Zn metal batteries

The reversibility and stability of aqueous Zn metal batteries (AZMBs) are largely limited by Zn dendrites and interfacial parasitic reactions. Herein, we propose a parallel modulation strategy to boost the reversibility of the Zn anode by introducing N,N,N’,N’-tetramethylchloroformamidinium hexafluorophosphate (TCFH) as an additive in the electrolyte. TCFH is composed of PF6− and TN+ with opposite charges. PF6− can spontaneously induce the in-situ generation of ZnF2 solid electrolyte interface (SEI) on the anode, which can improve the transport kinetics of Zn2+ at the interface, thus promoting the rapid and uniform deposition of Zn as well as inhibiting the growth of dendrites. In addition, TN+ is enriched at the anode surface during Zn deposition through the anchoring effect, which brings a reconfiguration of the ion/molecule distribution. The anchored-TN+ reduces the concentrations of H2O and SO42−, sufficiently restraining the parasitic reaction. Thanks to the dual-phase interface engineering constructed of PF6− and TN+ in parallel, the symmetric cell with the proposed electrolyte survives long cycling stability over 750 h at 20 mA cm−2, 10 mAh cm−2. This study offers a distinct viewpoint to the multidimensional optimization of Zn anodes for high-performance AZMBs.

Local-global feature fusion network for hyperspectral image classification

ABSTRACT Hyperspectral images (HSIs) are rich in spatial and spectral information, making them crucial for accurate classification. While existing convolutional neural network (CNN) methods have shown promise in HSI classification, they primarily focus on local features and struggle with establishing long-range dependencies. To address these limitations, we propose a Local-Global Feature Fusion Network (LGFFN). LGFFN effectively learns joint local-global information from HSIs. Firstly, we employ a group-general convolution parallel module (GGCPM) to extract both local and global spectral features. Next, we utilize parallel branches: a CNN branch for capturing fine-grained spatial information and an improved transformer encoder (ITE) branch for capturing long-distance spatial-spectral features and exploiting semantic information. Finally, we elegantly integrate the diverse features obtained from these branches to obtain classification results. We validate our method on four datasets and compare it with six state-of-the-art hyperspectral classification methods. Our experimental results demonstrate that LGFFN achieves high classification accuracy, outperforming existing methods.

Optimization of steel plate quality inspection driven by PscSE and SPPFELAN

AbstractBased on the improved YOLOv8n, a steel plate defect detection and recognition method is proposed to address the high labor costs and workload of traditional tasks. SPPFELAN processes inputs in parallel to enhance computational efficiency by executing multiple pooling operations simultaneously. The parallel feature fusion module PscSE, using a mixed‐dimension SE attention mechanism (scSE), captures global and channel‐related information better, improving characterization capability. The EIOU loss function addresses the ambiguous aspect ratio definition of CIOU loss, enhancing detection accuracy and accelerating convergence. Results show the YOLOv8n‐PscSE‐SPPFELAN model achieves 76.9% mAP@0.5 on the Northeastern University steel plate dataset, a 4.6% improvement over the original YOLOv8n, with a computation amount of 7.7 GFLOPs, reducing resource usage and greatly improving detection speed.

병렬 운전을 통한 4kW급 인공위성용 아날로그 LLC 공진형 컨버터 개발

A modular analog LLC resonant converter capable of delivering high power for aerospace applications with zero voltage switching turn on of power semiconductors has been proposed. The LLC converter utilizes a half bridge configuration, with top and bottom switches designed by paralleling two low capacity power semiconductors suitable for use in space, thereby increasing the current capacity to enable an output of up to 1.2kW. Each LLC converter module takes in 100V power and outputs 300V in a step up DC/DC converter configuration, designed to operate with regulated voltage and current. This paper describes the operating principles and steady state analysis of the proposed DC/DC converter. To validate its effectiveness, a prototype was constructed and tested. The tests yielded a max efficiency of 95.3% and an output of approximately 4kW, demonstrating the feasibility and output performance of high efficiency, high power supplies through load sharing of multiple modules in parallel.

Enhancing Skin Disease Diagnosis with TFFNet: A Two-Stream Feature Fusion Network Integrating CNNs and Self Attention Block

The skin of an individual serves as the primary defense mechanism for safe guarding vital organs in the body. Although this barrier effectively protects internal organs from a variety of threats, it is still prone to damage from viral, fungal, or dust-related illnesses. Even minor skin injuries possess the potential to escalate into more severe and hazardous conditions. A prompt and precise skin disease diagnosis becomes crucial in expediting the healing process for individuals grappling with skin-related issues. The objective of this study is to develop a system based on Convolutional Neural Network (CNN) that can accurately identify various skin diseases. The proposed architecture, known as TFFNet (Two-Stream Feature Fusion Network), integrates two simultaneous modules featuring a Self-Attention (SA) block. We employ Self Attention-Convolutional Neural Networks (SACNNs) and Depthwise Separable Convolution (DWSC) to establish a diagnostic system for skin diseases. In this method, two separate CNN models are joined together, and two parallel modules (M1 and M2) are added. This greatly reduces the total number of trainable parameters. In comparison to other deep learning methods outlined in existing literature, the proposed CNN exhibits a notably lower number of learned parameters, specifically around 7 million for classification purposes. The skin disease classification was carried out on three datasets—ISIC2016, ISIC2017, and HAM10000. The model achieved testing accuracies of 89.70%, 90.52%, and 90.12% on each respective dataset.

Quarry rock reef design features influence fish assemblage structure across a systematically heterogenous restoration reef

To restore an area of temperate rocky-reef degraded by sedimentation, scour, and burial, a large quarry rock reef, the Palos Verdes Restoration Reef (PVR), was built with a heterogenous design including high relief elements intended to increase fish biomass productivity and support a diverse reef community. The replicated design features provide a unique opportunity to study the effects of reef design on fish habitat use patterns. To determine how submodule scale habitat features are associated with variation in the assemblage structure of eleven focal fish species on the PVR we conducted diver-operated stereo-video surveys on all 18 PVR modules 9–13 months after construction. The highest mean densities of most focal fish species and highest total fish densities were observed on high and medium-relief reef submodules and their adjacent ecotones positioned on the offshore sides or ends of modules. These included the most abundant species on the PVR, the zooplanktivorous Blacksmith (Chromis punctipinnis), as well as the fishery species Kelp Bass (Paralabrax clathratus) and California Sheephead (Bodianus pulcher). On the inshore side of parallel modules, the reef and ecotone transects on low and medium relief submodules exhibited the lowest total mean fish densities, and consistently lower mean focal fish species densities. Focal fish species assemblages also differed between the reef and sand-rock ecotone transects. Reef-resident planktivorous fishes likely contribute to reef primary and secondary productivity through consumer mediated nutrient transport and are an important consideration in restoration reef design. Future reef restoration designs should consider incorporating replicated heterogeneous design features including the placement of higher relief elements relative to shore and current patterns as a special consideration for providing habitat for planktivorous reef-resident fishes.