卷积增强自适应分类模型的构造与研究

doi:10.11896/jsjkx.241200069

摘要/Abstract

摘要： 经典卷积神经网络(Convolutional Neural Networks,CNN)已被成功应用于图像领域,但是在图像旋转与缩放等几何变换条件下提取图像特征存在鲁棒性不足的局限。文中提出一种卷积增强自适应分类模型(Convolutionally Enhanced Adaptive Classification Model,CEACM),通过集成特征提取与分类器优化,来提升模型在复杂图像变换场景下的性能。在特征提取部分,引入了特征不变层作为对传统CNN的增强机制。该层通过集成旋转变换策略,有效增强CNN在提取图像特征时的旋转不变性,确保模型能够从多样化的输入数据中捕获到稳定且具有高度代表性的特征表示,提高模型对图像几何变换的鲁棒性。在分类器设计部分,提出了一种基于粒子群优化(Particle Swarm Optimization,PSO)的自适应增强模型。该模型利用PSO算法的全局搜索能力,对分类器的权重进行精细调整,能有效避免传统优化方法易陷入局部最优解的问题,提升模型的泛化能力和分类精度。为验证CEACM模型的有效性,采用了一系列国际标准图像数据集进行测试。实验结果表明,相较于传统机器学习模型及现有改进CNN模型,CEACM在分类任务上展现出了更为优越的性能,不仅提高了分类的准确率,还显著增强了模型在处理图像几何变换时的稳定性与鲁棒性。

关键词: 数据增强, 卷积神经网络, 自适应增强算法, 粒子群优化算法

Abstract: Classical convolutional Neural Networks(CNNs) have been successfully widely used in the image application field.However,when images undergo rotations or scaling transformations,the relative positions and scales of features change,presenting challenges for traditional CNNs in extracting stable and invariant image features.To address this issue,this paper introduces a Con-volutional Enhanced Adaptive Classification Model(CEACM),which consists of two parts:feature extraction and classifier design.In the feature extraction stage,a feature invariant layer is used to enhance the CNN,applying rotational transformations to enhance the convolutional neural network features,allowing the model to extract stable and representative features from the input data.In the classifier part,an adaptive enhancement model based on Particle Swarm Optimization(PSO) is proposed,where the PSO algorithm is used to optimize the weights of the classifier to avoid local optima,thus improving the model’s generalization and classification performance.Finally,the model’s performance is evaluated using a series of image datasets.Experimental re-sults indicate that the proposed CEACM outperforms traditional machine learning models and a series of improved models in terms of classification effectiveness.

Key words: Data augmentation, Convolutional neural network, Adaptive boosting, Particle swarm optimization

中图分类号:

TP311

陈一卓, 邹伟, 王洪大. 卷积增强自适应分类模型的构造与研究[J]. 计算机科学, 2025, 52(11A): 241200069-5. https://doi.org/10.11896/jsjkx.241200069

CHEN Yizhuo, ZOU Wei, WANG Hongda. Construction and Research of Convolution Enhanced Adaptive Classification Model[J]. Computer Science, 2025, 52(11A): 241200069-5. https://doi.org/10.11896/jsjkx.241200069

参考文献

[1]PETERSEN P C,SEPLIARSKAIA A.VC dimensions of group convolutional neural networks [J].Neural Networks,2024,169:462-474.
[2]DAS H S,DAS A,NEOG A,et al.Breast cancer detection:Shallow convolutional neural network against deep con-volutional neural networks based approach [J].Frontiers in Genetics,2023,13:1097207.
[3]HUANG S Y,AN W J,ZHANG D S,et al.Image classification and adversarial robustness analysis based on hybrid quantum-classical convolutional neural network [J].Optics Communications,2023,533:129287.
[4]NGO G,BEARD R,CHANDRA R.Evolutionary bagging forensemble learning [J].Neurocomputing,2022,510:1-4.
[5]ZHU L,XU H,CUI X.Improved AdaBoost algorithm based on base classifier coefficients and diversity [J].Journal of Compu-ter Applications,2021,41(8):2225-2231.
[6]AN X,HU C,LI Z,et al.Decentralized AdaBoost algorithm over sensor networks [J].Neurocomputing,2022,479:37-46.
[7]BUSARI G A,LIM D H.Crude oil price prediction:A comparison between AdaBoost-LSTM and AdaBoost-GRU for improving forecasting performance [J].Computers & Chemical Engineering,2021,155:107513.
[8]MILOŠEVIĆ N,RACKOVIĆ M.Classification based on missing features in deep convolutional neural networks [J].Neural Network World,2019,29(4):221-234.
[9] MIRKHAN M,MEYBODI M R.Restricted convolutional neural networks [J].Neural Processing Letters,2019,50(2):1705-1733.
[10]SARIGÜL M,OZYILDIRIM B M,AVCI M.Differential convolutional neural network [J].Neural Networks,2019,116:279-287.
[11]WANG Z,YAO L.Expression Recognition Method Based on Convolutional Neural Network and Capsule Neural Network [J].Computers,Materials & Continua,2024,79(1):1-10.
[12]AKSOY M Ç,BERIL S,CEM Ü.Land classification in satellite images by injecting traditional features to CNN models [J].Remote Sensing Letters,2023,14(2):157-167.
[13]LI Y.Quantum AdaBoost algorithm via cluster state [J].International Journal of Modern Physics B,2017,31(6):1750040.
[14]LEE W,CHI-HYUCK J,LEE J S.Instance categorization bysupport vector machines to adjust weights in AdaBoost for imbalanced data classification [J].Information Sciences,2017,381:92-103.
[15]DING Y,ZHU H Y,CHEN R Y,et al.An efficient AdaBoost algorithm with the multiple thresholds classification [J].Applied Sciences,2022,12:5872.
[16]NIU X,MA W.Selective quantum ensemble learning inspired by improved AdaBoost based on local sample information [J].Complex & Intelligent Systems.2023,9(5):5173-5183.
[17]SAVARGIV M,BEHROOZ M,MOHAMMAD R K.A new ensemble learning method based on learning automata [J].Journal of Ambient Intelligence and Humanized Computing,2022,13(7):3467-3482.
[18]KAYED M,ANTER A,MOHAMED H.Classification of Garments from Fashion MNIST Dataset Using CNN LeNet-5 Architecture [C]//2020 International Conference on Innovative Trends in Communication and Computer Engineering(ITCE).Aswan,Egypt,2020:238-243.
[19]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks [C]//Communications of the ACM.2012:84-90.
[20]DING X,LI Q,CHENG Y,et al.Local keypoint-based Faster R-CNN [J].Applied Intelligence,2020,50:3007-3022.
[21]ROY D,CHAKRABORTY I,ROY K.Scaling Deep SpikingNeural Networks with Binary Stochastic Activations[C]//2019 IEEE International Conference on Cognitive Computing(ICCC).Milan,Italy,2019:50-58.
[22]HUANG X.Improved Model Based onGoogLeNet and Residual Neural Network ResNet[C]//International Journal of Cognitive Informatics and Natural Intelligence(IJCINI).2022:1-19.
[23]DONG Z,LIN S.Research on image classification based onCapsnet[C]//2019 IEEE 4th Advanced Information Technology,Electronic and Automation Control Conference(IAEAC).Chengdu,China,2019:1023-1026.
[24]OBAID K B,ZEEBAREE S R M,AHMED O M.Deep Learning Models Based on Image Classification:A Review[J].International Journal of Science and Business,2020,4.
[25]YAN Z,JAGADEESH V,DECOSTE D,et al.HD-CNN:Hierarchical Deep Convolutional Neural Network for Image Classification[C]//International Conference on Computer Vision(ICCV).2015,2:435-443.
[26]MANESSI F,ROZZA A.Learning combinations of activation functions[C]//2018 24th International Conference on Pattern Recognition(ICPR).IEEE,2018:61-66.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed