Small Kernel Research Articles

IMCMK-CNN: A lightweight convolutional neural network with Multi-scale Kernels for Image-based Malware Classification

Rapid and accurate identification of unknown malware and its variants is the premise and basis for the effective prevention of malicious attacks. However, with the explosive growth of malware variants, the efficiency of manual updating of the sample database is getting worse and worse. It is difficult for the traditional identification methods to effectively capture the sample feature information operated by the confusion method only based on the delayed database information. The research into the direction of malware detection is dedicated to surmounting the limitations of conventional detection methodologies, and delves deeply into the application of cutting-edge technologies such as data visualization, machine learning, and hybrid detection within the realm of malware detection. Through these investigations, our goal is to construct a detection system that is both more precise and efficient, capable of addressing the ever-evolving threats to cybersecurity. Pursuing research in this direction is not only vital for enhancing network security defenses and safeguarding user data, but it will also foster the advancement of related state-of-the-art technologies and further mitigate the economic and societal repercussions of malware attacks. In light of this issue, this paper proposes the Image-based Malware Classification with Multi-scale Kernels (IMCMK), a Convolutional Neural Network (CNN) architecture using multi-scale convolution kernels mixing action to improve malware variants detection capabilities. First, we propose the Multi-scale Kernels (MK) block combining deep large kernel convolution and standard small kernel convolution with shortcuts to improve the accuracy. Furthermore, we propose Multi-scale Kernel Fusion (MKF) to reduce the number of parameters that come with the large kernels. The improved Squeeze-and-Excitation (SE) block can obtain the correlation between different channels to further increase the model performance. Experimental results show that IMCMK outperforms the state-of-the-art methods in malware family classification accuracy, which has achieved 99.25 %.

Rolling bearing fault diagnosis method based on MSDCNN in strong noise environment

Aiming at the problems of poor noise resistance, high computational complexity and insufficient generalization performance of traditional bearing fault diagnosis methods based on deep learning, a rolling bearing fault diagnosis method based on multi-scale dynamic convolutional neural network (MSDCNN) is proposed. Firstly, the one-dimensional vibration signal of the rolling bearing is converted to the frequency domain by Fourier transform, and the features are further extracted by wide convolution kernel. Secondly, a multi-scale dynamic convolution structure is proposed, and the feature information extracted by convolution kernels of different sizes is given different weights by using an improved channel attention mechanism. Then, a self-calibrating spatial attention mechanism (SCSAM) is designed, and the extracted feature information is input into the spatial attention mechanism to capture the importance of different regions. Finally, the features are further extracted by small convolution kernels, and the fault category is classified by using Softmax classifier. The fault diagnosis performance of the proposed model is verified by three different data sets. The experimental results show that compared with other intelligent models, the proposed model has higher classification accuracy, better generalization ability and stronger robustness under strong noise background.

A novel embedded kernel CNN-PCFF algorithm for breast cancer pathological image classification

Early screening of breast cancer through image recognition technology can significantly increase the survival rate of patients. Therefore, breast cancer pathological image is of great significance for medical diagnosis and clinical research. In recent years, numerous deep learning models have been applied to breast cancer image classification, with deep CNN being a typical representative. Due to the use of multi-depth small convolutional kernels in mainstream CNN architectures such as VGG and Inception, the obtained image features often have high dimensionality. Although high dimensionality can bring more fine-grained features, it also increases the computational complexity of subsequent classifiers and may even lead to the curse of dimensionality and overfitting. To address these issues, a novel embedded kernel CNN principal component feature fusion (CNN-PCFF) algorithm is proposed. The constructed kernel function is embedded in the principal component analysis to form the multi-kernel principal component. Multi-kernel principal component analysis is used to fuse the high dimensional features obtained from the convolution base into some representative comprehensive variables, which are called kernel principal components, so as to achieve the purpose of dimensionality reduction. Any type of classifier can be added based on multi-kernel principal components. Through experimental analysis on two public breast cancer image datasets, the results show that the proposed algorithm can improve the performance of the current mainstream CNN architecture and subsequent classifiers. Therefore, the proposed algorithm in this paper is an effective tool for the classification of breast cancer pathological images.

Study on the Damage Characteristics of Wheat Kernels under Continuous Compression Conditions.

Peeling wheat yields higher-quality flour. During processing in a flaking machine, wheat kernels undergo continuous compression within the machine's chamber. As this compression persists, damage to the kernels intensifies and accumulates, eventually leading to kernel breakage. To study the damage characteristics of wheat kernels during peeling, this study established a continuous damage model based on Hertzian contact theory and continuous damage theory. The model's accuracy was validated through experiments, culminating in the calculation of critical parameters for wheat peeling. This study focused on different wheat varieties (Ningmai 22 and Jichun 1) and kernel sizes (the thicknesses of the small, medium, and large kernels were standardized as follows: Ningmai 22-2.67 ± 0.07 mm, 2.81 ± 0.07 mm, and 2.95 ± 0.07 mm; Jichun 1-2.98 ± 0.11 mm, 3.20 ± 0.11 mm, and 3.42 ± 0.11 mm). Continuous compression tests were conducted using a mass spectrometer, and critical damage parameters were analyzed and calculated by integrating the theoretical model with experimental data. The test results showed that the average maximum crushing force (Fc) for small, medium, and large-sized kernels of Ningmai 22 was 96.71 ± 2.27 N, 110.17 ± 2.68 N, and 128.41 ± 2.85 N, respectively. The average maximum crushing deformation (αc) was 0.65 ± 0.08 mm, 0.68 ± 0.13 mm, and 0.77 ± 0.17 mm, respectively. The average elastic-plastic critical pressure (Fs) was 50.21 N, 60.13 N, and 59.08 N, respectively, and the average critical values of elastic-plastic deformation (αs) were 0.37 mm, 0.38 mm, and 0.39 mm, respectively. For Jichun 1, the average maximum crushing force (Fc) for small-, medium-, and large-sized kernels was 113.34 ± 3.15 N, 125.28 ± 3.64 N, and 136.15 ± 3.29 N, respectively. The average maximum crushing deformation (αc) was 0.75 ± 0.11 mm, 0.83 ± 0.15 mm, and 0.88 ± 0.18 mm, respectively. The average elastic-plastic critical pressure (Fs) was 58.11 N, 64.17 N, and 85.05 N, respectively, and the average critical values of elastic-plastic deformation (αs) were 0.45 mm, 0.47 mm, and 0.52 mm, respectively. The test results indicated that during mechanical compression, if the deformation is less than αs, the continued application of the compression load will not result in kernel crushing. However, if the deformation exceeds αs, continued compression will lead to kernel crushing, with the required number of compressions decreasing as the deformation increases. If the deformation surpasses αc, a single compression load is sufficient to cause kernel crushing. Since smaller wheat kernels are more susceptible to breakage during processing, the peeling pressure (F) within the chamber should be controlled to remain below the average elastic-plastic critical pressure (Fs) of small-sized wheat kernels. Additionally, the kernel deformation (α) induced by the flow rate and loading in the chamber should be kept below the average elastic-plastic critical deformation (αs) of small-sized wheat kernels. This paper provides a theoretical foundation for the structural design and optimization of processing parameters for wheat peeling machines.

A consistent model of the initiation, early expansion, and possible extinction of a spark-ignited flame kernel

Modelling the establishment and growth of spark-ignited (SI) flame kernels has always been a topic of great interest, especially due to their key role in affecting the performance of SI engines. A major issue is that the unsteady conditions and the small kernel size hinder the application of the typical (both linear and non-linear) flame stretch correlations, valid only long after the ignition stage. Overcoming such limitations, this work presents a novel, mathematically consistent, and compact model that enables prediction of flame kernel initiation and early expansion, including its possible extinction. Firstly, spark-driven initiation models from literature are discussed, and an effective flame kernel initiation method is proposed. Then, the expansion model is defined complementing the mass, energy, and species conservation equations for the spherical kernel with the reactant and temperature profiles outside of it using the theory of transient thermodiffusive flames. After accounting for the convective flow caused by the combustion-induced density reduction and the variable thermodynamic properties of the reacting fuel/air mixture, the result is a two-equation model that predicts the kernel expansion even up to its possible extinction due to flame stretch. After calibration of the expansion model, successful validation is achieved against literature data on lean propane/air flames, and the influence of the model parameters is examined in detail. The proposed expansion model is formulated also aiming for inclusion into the simulation of combustion in SI engines, enabling more accurate predictions at part loads, as well as more effective estimation of the cycle-to-cycle variation thanks to the good model sensitivity to the parameters most affecting the ignition.

Fusion of Multiple Attention Mechanisms and Background Feature Adaptive Update Strategies in Siamese Networks for Single-Object Tracking

Single-object tracking algorithms based on Siamese full convolutional networks have attracted much attention from researchers owing to their improvement in precision and speed. Since this tracking model only learns a similarity model offline, it is not able to obtain more useful feature discrimination information to adapt to the various variations of targets in complex scenes. To improve the performance of this tracking model, we propose a Siamese network tracking algorithm that incorporates multiple attention mechanisms and an adaptive updating strategy for background features. First, a backbone feature extraction network is proposed that utilizes a small convolutional kernel to fuse jump-layer connectivity features, thereby improving the feature representation capability of the network. Second, an adaptive update strategy for background features is proposed to improve the model’s ability to discriminate between the object and background features. Third, the fusion of multiple attention mechanisms is proposed so that the model learns to focus on the channel, spatial, and coordinate features. Fourth, the response fusion operation is proposed after the inter-correlation operation to enrich the output response of the model. Finally, our algorithm is trained using the GOT-10K dataset and evaluated by testing on the object tracking benchmark datasets OTB100 and VOT2018. The test results show that compared with other algorithms, our algorithm can effectively cope with the problem of degradation of the tracking performance in complex environments, and it can further improve the tracking precision and precision under the premise of ensuring the tracking speed.

GC[formula omitted]: Grouped convolutional color constancy

Computational color constancy is an ill-posed problem since different combinations of objects and illumination may appear the same color in an image. Most methods produce a single illuminant estimate and cannot explain the potential ambiguity between possible illuminant solutions. In this work, a grouped convolutional color constancy (GC3) method is proposed. It approaches the computational color constancy as a grouped regression problem, dealing with ambiguity by generating multiple possible illuminant solutions. The GC3 consists of two components: fast feature extraction (FFE) and grouped illuminant regression (GIR). FFE uses small convolution kernels and rapid down-sampling scheme for quick feature extraction, while GIR employs grouped convolution layers for feature grouping regression, which estimates the illuminant solution and its probability for each group with two separate branches. The final illumination estimation is obtained by averaging all illuminant solutions, weighted by the normalized probability for each group. The proposed method yields competitive results on the ColorChecker dataset and achieves the state-of-the-art accuracy on the NUS and Cube+ datasets. Moreover, when compared with the state-of-the-art methods, our proposed approach exhibits superior time efficiency, making it the most efficient method available.

ASF-LKUNet: Adjacent-scale fusion U-Net with large kernel for multi-organ segmentation

In the multi-organ segmentation task of medical images, there are some challenging issues such as the complex background, blurred boundaries between organs, and the larger scale difference in volume. Due to the local receptive fields of conventional convolution operations, it is difficult to obtain desirable results by directly using them for multi-organ segmentation. While Transformer-based models have global information, there is a significant dependency on hardware because of the high computational demands. Meanwhile, the depthwise convolution with large kernel can capture global information and have less computational requirements. Therefore, to leverage the large receptive field and reduce model complexity, we propose a novel CNN-based approach, namely adjacent-scale fusion U-Net with large kernel (ASF-LKUNet) for multi-organ segmentation. We utilize a u-shaped encoder–decoder as the base architecture of ASF-LKUNet. In the encoder path, we design the large kernel residual block, which combines the large and small kernels and can simultaneously capture the global and local features. Furthermore, for the first time, we propose an adjacent-scale fusion and large kernel GRN channel attention that incorporates the low-level details with the high-level semantics by the adjacent-scale feature and then adaptively focuses on the more global and meaningful channel information. Extensive experiments and interpretability analysis are made on the Synapse multi-organ dataset (Synapse) and the ACDC cardiac multi-structure dataset (ACDC). Our proposed ASF-LKUNet achieves 88.41% and 89.45% DSC scores on the Synapse and ACDC datasets, respectively, with 17.96M parameters and 29.14 GFLOPs. These results show that our method achieves superior performance with favorable lower complexity against ten competing approaches.ASF-LKUNet is superior to various competing methods and has less model complexity. Code and the trained models have been released on GitHub.

A novel method for the classification of 3D point clouds based on the improved PointNet++

Abstract In deep learning, point clouds are used as the primary input format for 3D data, which can provide detailed geometric information about objects in the original 3D space. PointNet++ is a deep learning network that uses point cloud data as an input format, which avoids the losses associated with the previous conversion of point cloud into 3D voxelization and a collection of 2D images. Although PointNet++ can directly process point cloud data in various ways, due to the disordered, irregular, and unevenly distributed nature of point cloud data, the effect of extracting point cloud features could be better. The large amount of point cloud data also leads to the training model falling into the local optimal solution, which affects the training results. In recent years, some effective methods and strategies have emerged to address these problems. In this thesis, three methods are proposed based on the PointNet++ network: feature similarity-based attention pooling, small kernel convolution, and diverse branch block method to improve the performance of the PointNet++ network. Experiments show that the improvement methods proposed in this paper effectively improve the feature extraction accuracy, which improves the accuracy of the PointNet++ network for classification on the ModelNet40_Normal_Resampled dataset, with an overall improvement of 1% compared with PointNet++.

SSL-LRN: A Lightweight Semi-Supervised-Learning-Based Approach for UWA Modulation Recognition

Due to the lack of sufficient valid labeled data and severe channel fading, the recognition of various underwater acoustic (UWA) communication modulation types still faces significant challenges. In this paper, we propose a lightweight UWA communication type recognition network based on semi-supervised learning, named the SSL-LRN. In the SSL-LRN, a mean teacher–student mechanism is developed to improve learning performance by averaging the weights of multiple models, thereby improving recognition accuracy for insufficiently labeled data. The SSL-LRN employs techniques such as quantization and small convolutional kernels to reduce floating-point operations (FLOPs), enabling its deployment on underwater mobile nodes. To mitigate the performance loss caused by quantization, the SSL-LRN adopts a channel expansion module to optimize the neuron distribution. It also employs an attention mechanism to enhance the recognition robustness for frequency-selective-fading channels. Pool and lake experiments demonstrate that the framework effectively recognizes most modulation types, achieving a more than 5% increase in recognition accuracy at a 0 dB signal-to-noise ratio (SNRs) while reducing FLOPs by 84.9% compared with baseline algorithms. Even with only 10% labeled data, the performance of the SSL-LRN approaches that of the fully supervised LRN algorithm.

Toward compound fault diagnosis via EMAGAN and large kernel augmented few-shot learning

Bearings are essential in machinery. Damage to them can cause financial losses and safety risks at industrial sites. Therefore, it is necessary to design an accurate diagnostic model. Although many bearing fault diagnosis methods have been proposed recently, they still cannot meet the requirements of high-accurate prediction of bearing faults. There are several challenges in this: 1) In practical settings, gathering sufficient and balanced sample data for training diagnostic network models proves challenging. 2) The damage to bearings in real industrial production sites is not singular, and compound faults are also a huge challenge for diagnostic networks. To address these issues, this study introduces a novel fault diagnosis model called EMALKNet that integrates DCGAN with Efficient Multi-Scale Attention (EMAGAN) and RepLKNet-XL, enhancing the detection and analysis of bearing faults in industrial machinery. This model employs EMAGAN to explore the underlying distribution of raw data, thereby enlarging the fault sample pool and enhancing the model’s diagnostic capabilities; The large kernel structure of RepLKNet-XL is different from the current mainstream small kernel and has stronger representation extraction ability. The proposed method has been validated on the Paderborn University dataset and the Huazhong University of Science and Technology dataset.

Rethinking encoder-decoder architecture using vision transformer for colorectal polyp and surgical instruments segmentation

Accurate polyp segmentation from colonoscopy images is important for the immediate diagnosis and effective treatment of colon cancer. While significant progress has been made in the polyps segmentation task, there are various challenges that need to be addressed. Polyps can vary greatly in size and shape, and often has no clear boundary between surrounding tissues. Furthermore, surgical instrument segmentation can aid surgeons with the precise positioning and orientation of the instruments, helping them to plan the next steps in the robot-assisted surgery. The proposed colorectal polyp segmentation Transformer (CPS-Former) uses innovative attention blocks in a network that encodes-decodes features like classic Semantic Segmentation Network (SegNet). However, it has special self-attention modules with small convolutional kernels that efficiently extract information from different feature-channels. Moreover, it is equipped with an effective positional embedding to capture information from a large area of context for long distance interactions. Additionally, a fusion block is embedded for scaling-attention that combines the outputs from the encoder-decoder blocks to enhance the semantic features and reduce the non-semantic ones. Transformer encoder blocks also modified by adding a local feedforward layer and skips connections, and adjust the channel sizes to reduce the model trainable parameters. We evaluate our colorectal polyp segmentation network (CPS-Former) on four colorectal polyp public datasets and one surgical instrument segmentation dataset, which show its superiority over other state-of-the-art polyp segmentation models. Our implementation source code and network weights are available at GitHub: https://github.com/ahmedeqbal/CPS-Former.

Dual residual and large receptive field network for lightweight image super-resolution

In recent years, there has been rapid progress in single-image super-resolution (SISR) reconstruction technology based on deep learning, but many methods face challenges in practical application due to their excessive computational requirements. To address this issue, numerous lightweight super-resolution (SR) methods have been proposed. Convolutional neural networks (CNNs) have demonstrated outstanding performance in lightweight SR reconstruction tasks. However, existing CNN-based lightweight SR networks still suffer from redundancy due to the use of small convolutional kernels and deeper network architectures for feature extraction. To solve these problems, we propose a novel network called Dual Residual and Large Receptive Field Network (DRLN), which includes two main designs. Firstly, a Large Receptive Field Feature Extraction Block (LRFEB) is proposed, which concatenates large kernel depth separable convolutions and large kernel depth separable dilated convolutions, effectively capturing spatial contextual information without increasing the number of parameters or computational burden, and enhancing the overall performance of the model. Secondly, in the Efficient Feature Distillation Block (EFDB), we innovatively propose a dual residual structure, which greatly improve the learning ability and feature expression richness of the network, and significantly improve the reconstruction performance. Extensive experiments demonstrate that DRLN surpasses BSRN which is the champion of the model complexity track of the NTIRE 2022 Efficient SR Challenge on all benchmark datasets, with fewer parameters and less computation.

The Infection of Barley at Different Growth Stages by Bipolaris sorokiniana and Its Effect on Plant Yield and Sowing Value

Bipolaris sorokiniana, a barley pathogen, can infect via seeds, co-growing plants, or soil, causing yield and seed value reductions. This study aimed to assess B. sorokiniana’s impact at various growth stages on seed yield and quality. Results showed no direct impact on yield (which ranged from 0.49 kg·m−2 for the naked variety Rastik to 0.77 kg·m−2 for the Widawa variety), but revealed significant seed quality differences. Thousand kernel weight (TKW) and germination capacity (GC) varied notably between examined varieties. The largest kernels were observed for the Bryl variety (27.33 g), which also had the best germination (82.8%). The variety Ryton had the smallest kernels (24.04 g), while the poorest germination (56.7%) was observed for naked kernels of Rastik. A seed health analysis found a relation between inoculation stage and the intensity of spontaneous infection by pathogenic fungi, ranging from 86.4% for the control to more than 95% for the kernels harvested from plants inoculated at the flowering stage. Strong correlation emerged between B. sorokiniana grain infection and germination capacity, highlighting the fungus’s role in seed quality decline.

Aerial Image Segmentation of Nematode-Affected Pine Trees with U-Net Convolutional Neural Network

Pine wood nematode disease, commonly referred to as pine wilt, poses a grave threat to forest health, leading to profound ecological and economic impacts. Originating from the pine wood nematode, this disease not only causes the demise of pine trees but also casts a long shadow over the entire forest ecosystem. The accurate identification of infected trees stands as a pivotal initial step in developing effective prevention and control measures for pine wilt. Nevertheless, existing identification methods face challenges in precisely determining the disease status of individual pine trees, impeding early detection and efficient intervention. In this study, we leverage the capabilities of unmanned aerial vehicle (UAV) remote sensing technology and integrate the VGG classical small convolutional kernel network with U-Net to detect diseased pine trees. This cutting-edge approach captures the spatial and characteristic intricacies of infected trees, converting them into high-dimensional features through multiple convolutions within the VGG network. This method significantly reduces the parameter count while enhancing the sensing range. The results obtained from our validation set are remarkably promising, achieving a Mean Intersection over Union (MIoU) of 81.62%, a Mean Pixel Accuracy (MPA) of 85.13%, an Accuracy of 99.13%, and an F1 Score of 88.50%. These figures surpass those obtained using other methods such as ResNet50 and DeepLab v3+. The methodology presented in this research facilitates rapid and accurate monitoring of pine trees infected with nematodes, offering invaluable technical assistance in the prevention and management of pine wilt disease.

Robust DOA Estimation Using Multi-Scale Fusion Network with Attention Mask

To overcome the limitations of traditional methods in reverberant and noisy environments, a robust multi-scale fusion neural network with attention mask is designed to improve direction-of-arrival (DOA) estimation accuracy for acoustic sources. It combines the benefits of deep learning and complex-valued operations to effectively deal with the interference of reverberation and noise in speech signals. The unique properties of complex-valued signals are exploited to fully capture inherent features and rich information is preserved in the complex field. An attention mask module is designed to generate distinct masks for selectively focusing and masking based on the input. After that, the multi-scale fusion block efficiently captures multi-scale spatial features by stacking complex-valued convolutional layers with small size kernels, and reduces the module complexity through special branching operations. Experimental results demonstrate that the model achieves significant improvements over other methods for speaker localization in reverberant and noisy environments. It provides a new solution for DOA estimation for acoustic sources in different scenarios, which has significant theoretical and practical implications.

MSCLK: Multi-scale fully separable convolution neural network with large kernels for early diagnosis of Alzheimer’s disease

Alzheimer’s disease (AD) is identified as a central nervous system disease that exhibits irreversible degeneration, while mild cognitive impairment (MCI) is viewed as the preliminary stage of AD, and its pathogenesis is notably intricate. MCI contains two stages: early MCI (EMCI), and late MCI (LMCI). EMCI diagnosis can prevent EMCI from progressing to LMCI, and then to AD. Therefore, accurate diagnosis of EMCI/LMCI is crucial for developing the early intervention and treatment strategies of AD. Currently, most existing EMCI/LMCI diagnostic methods use single modality images, while different modality images carry different complementary information that helps for accurate diagnosis of EMCI/LMCI, and the lesion area is usually not limited to a single brain area, which involves multiple regions. In this case, conventional convolution operations cannot be able to accurately extract the pathological features of AD. In this work, we propose a novel Multi-scale fully Separable Convolution neural network with Large Kernels (MSCLK) method to diagnose early Alzheimer’s disease with structural Magnetic Resonance Imaging (sMRI) images. MSCLK mainly consists of the multi-scale 3D fully separable convolution modules and the deep metric learning module. The multi-scale convolution that contains both small and large kernels is used to effectively capture the discrimination features of different scale acceptance domains. 3D fully separable convolution is used to reduce parameters and overfitting. The deep metric learning is used to learn hard samples that are similar but belong to different classes. We also propose a variant method of MSCLK (called MSCLK-Fusion MRI and PET, MSCLK-FMP) by adding the pixel-level fusion module and feature-level fusion module into the MSCLK framework to integrate the sMRI image and the Positron Emission Computed Tomography (PET) image for further improving the accuracy of EMCI vs. LMCI classification task. The pixel-level fusion is used to achieve early pixel-level fusion of sMRI and PET images, and the feature-level fusion is used to achieve high-dimensional feature-level fusion of sMRI and PET images. Experimental results on the ADNI database show that the performance of our MSCLK and MSCLK-FMP are superior to other state-of-the-art methods. The accuracy of MSCLK achieves 98.89%, 95.97%, 96.39% and 98.76% for AD vs. EMCI, AD vs. LMCI, EMCI vs. NC and LMCI vs. NC classification tasks, respectively, and MSCLK-FMP achieves 93.93% for EMCI vs. LMCI classification task, indicating that MSCLK/MSCLK-FMP can be effectively used for diagnosing MCI patients. Moreover, our MSCLK-FMP is capable of pinpointing key brain areas involved in the pathological progression of MCI, such as the Temporal_Inf, the Hippocampus, the Precuneus, the Precentral, and the Thalamus. These findings contribute to uncovering the early onset of AD pathogenesis.

Graph-based minimum error entropy Kalman filtering

When the Gaussian kernel function is chosen with a small kernel bandwidths, the minimum error entropy Kalman filter (MEEKF) exhibits excellent performance. However, further narrowing the kernel bandwidths leads to a degradation in performance. To address this issue, this paper proposed a Graph-based MEEKF (G-MEEKF) algorithm based on Graph Signal Processing (GSP) theory. The G-MEEKF algorithm addresses the problem by smoothing the errors using graph filtering and incorporating graph topology into the cost function. The convergence of the algorithm is theoretically analyzed and simulations demonstrate that G-MEEKF improves the performance of MEEKF under small kernel bandwidths. Furthermore, it is emphasized that the topology of the graph abstracted from the minimum error entropy (MEE) definition also influences the performance of the algorithm.

Global Receptive Field Designed Complex-Valued Convolutional Neural Network Equalizer for Optical Fiber Communications

In this paper, an improved complex-valued convolutional neural network (CvCNN) structure to be placed at the received side is proposed for nonlinearity compensation in a coherent optical system. This complex-valued global convolutional kernel-assisted convolutional neural network equalizer (CvGNN) has been verified in terms of Q-factor performance and complexity compared to seven other related nonlinear equalizers based on both the 64 QAM experimental platform and the QPSK numerical platform. The global convolution operation of the proposed CvGNN is more suitable for the calculation process of perturbation coefficients, and the global receptive field can also be more effective at extracting effective information from perturbation feature maps. The introduction of CvCNN can directly focus on the complex-valued perturbation feature maps themselves without separately processing the real and imaginary parts, which is more in line with the waveform-dependent physical characteristics of optical signals. Based on the experimental platform, compared with the real-valued neural network with small convolutional kernel (RvCNNC), the proposed CvGNNC improves the Q-factor by ∼2.95 dB at the optimal transmission power, while reducing the time complexity by ∼44.7%.

Image Super-resolution Networks based on Structure Reparameterized Convolution

The traditional image super resolution method based on convolutional neural network generally brings two problems: First, the model has only a single scale of receptive field, which cannot use the image information in a smaller range; Second, the size of the feature map will become smaller and smaller in the process of continuous convolution, and it is necessary to continuously carry out edge zeroing operation to maintain the original size, which leads to the loss of part of the edge information. To solve the above two problems, we propose a network structure based on structure re-parameterized convolution, which sets a small convolution kernel in parallel beside a large convolution kernel in the same layer, trains the two cores simultaneously, and finally merges the two cores. The experimental results show that in this way, we make the large convolution kernel could capture smaller information, which not only improves the high-frequency details of the reconstructed image but also avoids frequent edge filling to reduce the information density, and effectively speeds up the reasoning speed after the model is re-parameterized. Compared with the advanced methods, our method achieves good results.