Asymmetric Network Research Articles

CellSAM: Advancing Pathologic Image Cell Segmentation via Asymmetric Large-Scale Vision Model Feature Distillation Aggregation Network.

Segment anything model (SAM) has attracted extensive interest as a potent large-scale image segmentation model, with prior efforts adapting it for use in medical imaging. However, the precise segmentation of cell nucleus instances remains a formidable challenge in computational pathology, given substantial morphological variations and the dense clustering of nuclei with unclear boundaries. This study presents an innovative cell segmentation algorithm named CellSAM. CellSAM has the potential to improve the effectiveness and precision of disease identification and therapy planning. As a variant of SAM, CellSAM integrates dual-image encoders and employs techniques such as knowledge distillation and mask fusion. This innovative model exhibits promising capabilities in capturing intricate cell structures and ensuring adaptability in resource-constrained scenarios. The experimental results indicate that this structure effectively enhances the quality and precision of cell segmentation. Remarkably, CellSAM demonstrates outstanding results even with minimal training data. In the evaluation of particular cell segmentation tasks, extensive comparative analyzes show that CellSAM outperforms both general fundamental models and state-of-the-art (SOTA) task-specific models. Comprehensive evaluation metrics yield scores of 0.884, 0.876, and 0.768 for mean accuracy, recall, and precision respectively. Extensive experiments show that CellSAM excels in capturing subtle details and complex structures and is capable of segmenting cells in images accurately. Additionally, CellSAM demonstrates excellent performance on clinical data, indicating its potential for robust applications in treatment planning and disease diagnosis, thereby further improving the efficiency of computer-aided medicine.

Multi-Modal Object Detection Method Based on Dual-Branch Asymmetric Attention Backbone and Feature Fusion Pyramid Network

With the simultaneous acquisition of the infrared and optical remote sensing images of the same target becoming increasingly easy, using multi-modal data for high-performance object detection has become a research focus. In remote sensing multi-modal data, infrared images lack color information, it is hard to detect difficult targets with low contrast, and optical images are easily affected by illuminance. One of the most effective ways to solve this problem is to integrate multi-modal images for high-performance object detection. The challenge of fusion object detection lies in how to fully integrate multi-modal image features with significant modal differences and avoid introducing interference information while taking advantage of complementary advantages. To solve these problems, a new multi-modal fusion object detection method is proposed. In this paper, the method is improved in terms of two aspects: firstly, a new dual-branch asymmetric attention backbone network (DAAB) is designed, which uses a semantic information supplement module (SISM) and a detail information supplement module (DISM) to supplement and enhance infrared and RGB image information, respectively. Secondly, we propose a feature fusion pyramid network (FFPN), which uses a Transformer-like strategy to carry out multi-modal feature fusion and suppress features that are not conducive to fusion during the fusion process. This method is a state-of-the-art process for both FLIR-aligned and DroneVehicle datasets. Experiments show that this method has strong competitiveness and generalization performance.

Adaptive Deep Ant Colony Optimization–Asymmetric Strategy Network Twin Delayed Deep Deterministic Policy Gradient Algorithm: Path Planning for Mobile Robots in Dynamic Environments

Path planning is one of the main focal points and challenges in mobile robotics research. Traditional ant colony optimization (ACO) algorithms encounter issues such as low efficiency, slow convergence, and a tendency to become stuck in local optima and search stagnation when applied to complex dynamic environments. Addressing these challenges, this study introduces an adaptive deep ant colony optimization (ADACO) algorithm, which significantly improves efficiency and convergence speed through enhanced pheromone diffusion mechanisms and updating strategies, applied to global path planning. To adapt to dynamically changing environments and achieve more precise local path planning, an asymmetric strategy network TD3 algorithm (ATD3) is further proposed, which utilizes global path planning information within the strategy network only, creating a new hierarchical path planning algorithm—ADACO-ATD3. Simulation experiments demonstrate that the proposed algorithm significantly outperforms in terms of path length and number of iterations, effectively enhancing the mobile robot’s path planning performance in complex dynamic environments.

Catch the Cyber Thief: A Multi-Dimensional Asymmetric Network Attack–Defense Game

This paper presents a novel multi-dimensional asymmetric game model for network attack–defense decision-making, called “Catch the Cyber Thief”. The model is built upon the concept of partially observable stochastic games (POSG) and is designed to systematically incorporate multi-dimensional asymmetry into network attack–defense problems. The attack agent is called a “thief” who wants to control a key host by exploring the unfamiliar network environment, and the defense agent is called a “police” who needs to catch the opponent before its goal is accomplished. The results indicate that the asymmetry of network attack and defense is not only influenced by attack and defense strategies but also by spatio-temporal factors such as the attacker’s initial position, network topology, and defense lag time. In addition, we have found that there may exist the “attack rhythm,” which makes “how to maintain a good attack rhythm” and “how to generate a robust defense strategy against different attackers” worth exploring. Compared with existing attack–defense game models, our game model can better generate a direct mapping relationship with real elements, enabling us to understand network attack and defense interactions better, recognize security risks, and design defense strategies that can directly serve real-world decision-making.

Brand livestream mode options: AI vs KOL livestream in the platform supply chain

PurposeDriven by the rapid expansion of online retail and the surge in livestream commerce, the impact of different livestream mode on brand and platform performance has become a critical issue. This paper analyzes the impact of artificial intelligence (AI) and key opinion leader (KOL) livestream on the profitability of brands and the platform, incorporating the effects of horizontal interactions to identify the optimal livestream mode.Design/methodology/approachThis paper develops a model of a platform supply chain involving two brands and a platform, where each brand independently decides whether to utilize KOL or AI livestream. Applying Stackelberg game approach, the study derives equilibria for various livestream scenarios, identifying the optimal livestream mode for both parties. Additionally, the model is extended to incorporate asymmetric market potential and network externality to evaluate their impact on a brand’s choice of livestream mode.FindingsSeveral interesting and important results are derived in this paper. Firstly, it is found that AI livestream enables brands to leverage network externality and mitigate the market disadvantage, thereby gaining a competitive advantage. Secondly, while KOL livestream promotes trust, the medium KOL commission rates could cause brands to be trapped in a prisoner’s dilemma, and excessively high commission rates may render them less profitable. Thirdly, the KOL commission rate, network externality intensity, horizontal interactions and market disadvantage are critical determinants influencing a brand’s choice of livestream mode.Originality/valueThis study is the first to investigate the effects of horizontal interactions, asymmetric market potential and asymmetric network externality on livestream mode selection by brands within a platform supply chain. The research provides valuable insights into optimizing livestream strategies to enhance brand profitability.

Asymmetric sandwich poly(lactic acid) composite films fabricated by one-step phase separation approach towards absorption-dominant electromagnetic interference shielding and hydrophobicity

The absorption-dominant electromagnetic interference (EMI) shielding polymer composites with excellent hydrophobicity and environmental friendliness are highly demanded to alleviate secondary electromagnetic radiation in high-humidity surroundings. Herein, cobalt–nickel@carbon nanotubes-polylactic acid-silver nanowires (CoNi@CNTs-PLA-AgNWs) films with asymmetrical sandwich architectures were successfully fabricated via a one-step magnetic field-assisted non-solvent-induced phase separation methodology. AgNWs are mainly spread on the rough top layer, but CoNi@CNTs are primarily distributed in the porous bottom layer of CoNi@CNTs-PLA-AgNWs composite films. Owing to the unique asymmetrical magnetic-electric dual networks, 150CoNi@CNTs-PLA-7AgNWs film displays the fantastic averaged EMI SET of 53.7 dB and optimal A of 0.61, illuminating the dominance of absorption in EMI shielding and efficiently alleviating the secondary electromagnetic pollution. In order to enhance the stability of EMI shielding in damp environment, CoNi@CNTs-PLA-AgNWs composite films are dip-coated in 1H,1H,2H,2H-perfluorodecyltrimethoxysilane (FDTS) solution for hydrophobic modification. The excellent water repellency is achieved on both top and bottom surface of CoNi@CNTs-PLA-AgNWs films with WCA > 150° and > 120°, respectively, while maintaining the superior EMI shielding performance. Such facile one-step asymmetrical sandwich structure design strategy provides valuable insight into preparing and controlling hydrophobic absorption-dominant EMI shielding composites, which is anticipated to have numerous implications for smart and portable electronic devices in high-humidity environment.

MSAACNN for intense noise suppression in DAS-VSP records

Distributed optical fiber sensor (DAS) is an emerging acquisition technique and has begun to be widely applied in seismic exploration owing to its advantages in acquisition and deployment. Nonetheless, DAS record has a low signal-to-noise ratio (SNR) due to the intense background noise. How to suppress the DAS background noise and increase the SNR of the DAS records has gradually become one of the hot issues in the field of seismic data processing. To solve the challenging tasks in intense DAS noise suppression, a multiscale sparse asymmetric attention convolutional neural network (MSAACNN) is proposed. The network uses dilated convolutions to expand the receptive field and map more feature information. Moreover, asymmetric convolutions are introduced to form an asymmetric unit, aiming to strengthen the feature extraction ability and realize the interaction between feature information of different scales. Finally, a pyramid attention module is used to enhance the primary features and improve the network denoising performance. The experimental results show that MSAACNN can effectively suppress the complex background noise in DAS records, compared with the traditional denoising methods and typical convolutional neural network (CNN) architecture. Additionally, the recovered signal components in the processing results are clear and complete, with significantly improved SNR.

Evolution of cooperation on dynamical asymmetrical networks with costs

Social networks denote the structure of interactions among individuals. Adjusting social relationships is a typical reaction among humans attempting to interact with altruistic partners. Theoretical investigations proved that dynamical network structures promote the evolution of cooperation. However, individuals react differently to diverse social partners and interactions. Furthermore, a large variety of potential costs of partner switching occur in real-world interactions. Establishing and maintaining different interactions cost differently. Based on these motivations, this investigation studies the interplay between the dynamics on networks and the dynamics of networks, which entangles the evolution of strategies and topology of adaptive multilayer networks whose structure is divided into a gaming layer for reaping payoff and a learning layer for spreading strategy. Individuals react differently to overlapped and un-overlapped partners whose gaming interactions and learning interactions are identical or not. And the effects of overlap levels and rewiring costs on the evolution of cooperation are explored. Simulation results demonstrate that increasing the ratio of rewiring as well as the overlap levels can enhance the evolution of cooperation significantly. However, both too frequent rewiring and too high overlap levels can result in the increment of isolated nodes, and thus provide a sanctuary for defectors to survive and make cooperators hardly to occupy the whole population. Moreover, it is found that rewiring costs have dramatically different impacts on the evolution of cooperation for different overlap levels. For a low overlap level, increasing costs suppress the evolution of cooperation. Yet a high overlap level makes increasing costs favor the cooperator to spread. Our results reveal the condition for the domination of cooperation with asymmetrical interactions, which may provide a potential way to understand the evolutionary cooperation in human society.

SiamATA: an asymmetric target-aware and frequency domain task-aware Siamese network for visual tracking

SiamATA: an asymmetric target-aware and frequency domain task-aware Siamese network for visual tracking

MAFNet: Multimodal Asymmetric Fusion Network for Radar Echo Extrapolation

Radar echo extrapolation (REE) is a crucial method for convective nowcasting, and current deep learning (DL)-based methods for REE have shown significant potential in severe weather forecasting tasks. Existing DL-based REE methods use extensive historical radar data to learn the evolution patterns of echoes, they tend to suffer from low accuracy. This is because data of radar modality face difficulty adequately representing the state of weather systems. Inspired by multimodal learning and traditional numerical weather prediction (NWP) methods, we propose a Multimodal Asymmetric Fusion Network (MAFNet) for REE, which uses data from radar modality to model echo evolution, and data from satellite and ground observation modalities to model the background field of weather systems, collectively guiding echo extrapolation. In the MAFNet, we first extract overall convective features through a global shared encoder (GSE), followed by two branches of local modality encoder (LME) and local correlation encoders (LCEs) that extract convective features from radar, satellite, and ground observation modalities. We employ an multimodal asymmetric fusion module (MAFM) to fuse multimodal features at different scales and feature levels, enhancing radar echo extrapolation performance. Additionally, to address the temporal resolution differences in multimodal data, we design a time alignment module based on dynamic time warping (DTW), which aligns multimodal feature sequences temporally. Experimental results demonstrate that compared to state-of-the-art (SOTA) models, the MAFNet achieves average improvements of 1.86% in CSI and 3.18% in HSS on the MeteoNet dataset, and average improvements of 4.84% in CSI and 2.38% in HSS on the RAIN-F dataset.

Asymmetric Network Combining CNN and Transformer for Building Extraction from Remote Sensing Images.

The accurate extraction of buildings from remote sensing images is crucial in fields such as 3D urban planning, disaster detection, and military reconnaissance. In recent years, models based on Transformer have performed well in global information processing and contextual relationship modeling, but suffer from high computational costs and insufficient ability to capture local information. In contrast, convolutional neural networks (CNNs) are very effective in extracting local features, but have a limited ability to process global information. In this paper, an asymmetric network (CTANet), which combines the advantages of CNN and Transformer, is proposed to achieve efficient extraction of buildings. Specifically, CTANet employs ConvNeXt as an encoder to extract features and combines it with an efficient bilateral hybrid attention transformer (BHAFormer) which is designed as a decoder. The BHAFormer establishes global dependencies from both texture edge features and background information perspectives to extract buildings more accurately while maintaining a low computational cost. Additionally, the multiscale mixed attention mechanism module (MSM-AMM) is introduced to learn the multiscale semantic information and channel representations of the encoder features to reduce noise interference and compensate for the loss of information in the downsampling process. Experimental results show that the proposed model achieves the best F1-score (86.7%, 95.74%, and 90.52%) and IoU (76.52%, 91.84%, and 82.68%) compared to other state-of-the-art methods on the Massachusetts building dataset, the WHU building dataset, and the Inria aerial image labeling dataset.

A Channel-Sensing-Based Multipath Multihop Cooperative Transmission Mechanism for UE Aggregation in Asymmetric IoE Scenarios

With the continuous progress and development of technology, the Internet of Everything (IoE) is gradually becoming a research hotspot. More companies and research institutes are focusing on the connectivity and transmission between multiple devices in asymmetric networks, such as V2X, Industrial Internet of Things (IIoT), environmental monitoring, disaster management, agriculture, and so on. The number of devices and business volume of these applications have rapidly increased in recent years, which will lead to a large load of terminals and affect the transmission efficiency of IoE data transmission. To deal with this issue, it has been proposed to perform data transmission via multipath cooperative transmission with multihop transmission. This approach aims to improve transmission latency, energy consumption, reliability, and throughput. This paper designs a channel-sensing-based cooperative transmission mechanism (CSCTM) with hybrid automatic repeat request (HARQ) for user equipment (UE) aggregation mechanism in future asymmetric IoE scenarios, which ensures that IoE devices data can be transmitted quickly and reliably, and supports real-time data processing and analysis. The main contents of this proposed method include strategies of cooperative transmission and redundancy version (RV) determination, a joint combination of decoding process at the receiving side, and a design of transmission priority through ascending offset sort (AOS) algorithm based on channel sensing. In addition, multihop technology is designed for the multipath cooperative transmission strategy, which enables cooperative nodes (CN) to help UE to transmit data. As a result, it can be obtained that CSCTM provides significant advancements in latency and energy consumption for the whole system. It demonstrates improvements in enhanced coverage, improved reliability, and minimized latency.

Voxel completion and 3D asymmetrical convolution networks for Lidar semantic segmentation

Voxel completion and 3D asymmetrical convolution networks for Lidar semantic segmentation

CCM-Net: Color compensation and coordinate attention guided underwater image enhancement with multi-scale feature aggregation

Due to the light scattering and wavelength absorption in water, underwater images exhibit blurred details, low contrast, and color deviation. Existing underwater image enhancement methods are divided into traditional methods and deep learning-based methods. Traditional methods either rely on scene prior and lack robustness, or are not flexible enough resulting in poor enhancement effects. Deep learning methods have achieved good results in the field of underwater image enhancement due to their powerful feature representation ability. However, these methods cannot enhance underwater images with various degradations because they do not consider the inconsistent attenuation of different color channels and spatial regions. In this paper, we propose a novel asymmetric encoder-decoder network for underwater image enhancement, called CCM-Net. Concretely, we first introduce the prior knowledge-based encoder, which includes color compensation (CC) modules and feature extraction modules that consist of depth-wise separable convolution and global-local coordinate attention (GLCA). Then, we design a multi-scale feature aggregation (MFA) module to integrate shallow, middle, and deep features. Finally, we deploy a decoder to reconstruct the underwater images with the extracted features. Extensive experiments on publicly available datasets demonstrate that our CCM-Net effectively improves the visual quality of underwater images and achieves impressive performance.

Asym-UNet: An asymmetric U-shape Network for breast lesions ultrasound images segmentation

Precise segmentation of breast ultrasound images is essential for early breast cancer screening. However, the segmentation process is challenging due to the diverse morphology of tumors, blurred boundaries, and the similar grey intensity distribution between lesions and normal tissues. To overcome these obstacles, we propose the Asym-UNet, an asymmetric U-shaped network tailored for segmenting breast lesions with high accuracy and reliability. This network features a Multi-branch Residual Encoder (MRE), an External Attention Module (EAM), and a Boundary Detection Module (BDM). The MRE is designed to capture distinctive representations of breast lesions, facilitating accurate subsequent segmentation. The EAM directs the network to focus on the lesion areas while filtering out irrelevant information, thereby enhancing diagnostic precision. Furthermore, the BDM, incorporated to refine the boundaries of predicted masks, provides additional supervision and guidance through a multi-level prediction layer. We conducted evaluations of the Asym-UNet against thirteen other segmentation methods using two publicly available datasets. The results indicate that our network achieves state-of-the-art performance in breast tumor segmentation.

Failure mechanism of a tunnel in a soil–rock mixture based on a new combined finite–discrete element method

The mechanical properties and failure characteristics of soil–rock mixtures (SRMs) directly affect the stability of tunnels constructed in SRMs. A new SRM modelling method based on the combined finite–discrete element method (FDEM) was proposed. Using this new SRM modelling method based on the FDEM, the mechanical characteristics and failure behaviour of SRM samples under uniaxial compression, as well as the failure mechanism of SRMs around a tunnel, were further investigated. The study results support the following findings: (1) The modelling of SRM samples can be achieved using a heterogeneous rock modelling method based on the Weibull distribution. By adjusting the relevant parameters, such as the soil–rock boundaries, element sizes and modelling control points, SRM models with different rock contents and morphologies can be obtained. (2) The simulation results of uniaxial compression tests of SRM samples with different element sizes and morphologies validate the reliability and robustness of the new modelling method. In addition, with increasing rock content, VBP (volumetric block proportion), the uniaxial compressive strength and Young’s modulus increase exponentially, but the samples all undergo single shear failure within the soil or along the soil–rock interfaces, and the shear failure angles are all close to the theoretical values. (3) Tunnels in SRMs with different rock contents all exhibit X-shaped conjugate shear failure, but the fracture network propagation depth, the maximum displacement around the tunnel, and the failure degree of the tunnel in the SRM roughly decreases via a power function as the rock content increases. In addition, as the rock content increases, such as when VBP = 40%, large rocks have a significant blocking effect on fracture propagation, resulting in an asymmetric fracture network around the tunnel. (4) The comparisons of uniaxial compression and tunnel excavation simulation results with previous theoretical results, laboratory test results, and numerical simulation results verify the correctness of the new modelling method proposed in this paper.

Improving vertebral diagnosis in computed tomography scans: a clinically oriented attention-driven asymmetric convolution network for segmentation

ObjectiveVertebral segmentation in computed tomography (CT) images remains an essential issue in medical image analysis, stemming from the variability of vertebral shapes, high complex deformations, and the inherent ambiguity in CT scans. The purpose of this study is to develop advanced methods to effectively address this challenging task. MethodsWe proposed an attention-driven asymmetric convolution deep learning (AACDL) framework for vertebral segmentation. Specifically, our approach involves constructing a novel asymmetric convolutional U-shaped deep learning architecture to enhance feature extraction capabilities by increasing its depth for capturing richer spatial details. Further, we construct a pyramid global context module that integrates global context information through pyramid pooling to boost segmentation accuracy particularly in smaller anatomical regions. Sequential channel and spatial attention mechanisms are also implemented within the network to enable it to automatically concentrate on learning the most salient features and regions across different dimensions. ResultsThe performance precision of our network was rigorously assessed using a suite of four benchmark metrics: the Dice coefficient, mean intersection over union (mIoU), Precision rate, and F1-score. When compared against the ground truth (GT), our model delivered outstanding scores, attaining a Dice coefficient of 82.79%, an mIoU of 90.72%, a Precision rate of 90.19%, and an F1-score of 90.09%, each reflecting the commendable accuracy and reliability of our network's segmentation output. ConclusionThe proposed AACDL method successfully realizes accurate segmentation of vertebral CT images, thereby demonstrating significant potential for clinical applications with its robust performance metrics. Its ability to handle the complexities associated with vertebral segmentation paves the way for enhanced diagnostic and treatment planning processes in healthcare settings.

A Comprehensive Security Framework for Asymmetrical IoT Network Environments to Monitor and Classify Cyberattack via Machine Learning

The Internet of Things (IoT) is an important component of the smart environment, which produces a large volume of data that is considered challenging to handle. In addition, the IoT architecture is vulnerable to many cyberattacks that can target operational devices. Therefore, there is a need for monitoring IoT traffic to analyze, detect malicious activity, and classify cyberattack types. This research proposes a security framework to monitor asymmetrical network traffic in an IoT environment. The framework offers a network intrusion detection system (NIDS) to detect and classify cyberattacks, implemented using a machine learning (ML) model residing in the middleware layer of the IoT architecture. A dimensionality reduction technique known as principal component analysis (PCA) is utilized to facilitate data transmission, which is intended to be sent from the middleware layer to the cloud layer with reduced complexity and fewer unnecessary inputs without compromising the information content. Therefore, the reduced IoT traffic data are sent to the cloud and the PCA data are retransformed to approximate the original data for visualizing the IoT traffic. The NIDS is responsible for reporting the attack type to the cloud in the event of an attack. Our findings indicate that the proposed framework has promising results in classifying the attack type, which achieved a classification accuracy of 98%. In addition, the dimension of the IoT traffic data is reduced by around 50% and it has a similarity of around 90% compared to the original data.

Learning quadrupedal locomotion on tough terrain using an asymmetric terrain feature mining network

Learning quadrupedal locomotion on tough terrain using an asymmetric terrain feature mining network

LMANet: A Lightweight Asymmetric Semantic Segmentation Network Based on Multi-Scale Feature Extraction

LMANet: A Lightweight Asymmetric Semantic Segmentation Network Based on Multi-Scale Feature Extraction