基于迁移学习的军事少样本集成分类研究
Few-Shot Ensemble Classification of Military Images Based on Transfer Learning
DOI: 10.12677/csa.2024.148180, PDF,   
作者: 鲁磊纪, 余红霞, 肖红菊, 鲍 蕾*：陆军炮兵防空兵学院计算机教研室，安徽 合肥
关键词: 少样本图像分类；迁移学习；卷积神经网络；集成；Few-Shot Classification； Transfer Learning； ConvNets； Ensemble
摘要: 深度神经网络是一种需要大量的数据来进行有效训练的模型。军事装备类数据普遍存在数据量较少，无法满足深度神经网络的训练需求，容易出现过拟合的问题。针对该问题，本文引入迁移学习技术，通过构建多类型样本训练集，微调预训练模型，构建军事装备类集成分类器。实践证明迁移学习在少样本分类任务中的应用节省了模型训练时间，解决了模型过拟合及对数据标签依赖性强的问题，能有效提高基于深度学习的军事装备类小样本图像分类的准确性。
Abstract: A large amount of data is indispensable for effective training of deep neural networks. Military equipment data generally suffers from insufficient quantities, which fails to meet the training requirements of deep neural networks and easily leads to over fitting. To address this issue, this paper introduces transfer learning technology by constructing a multi-type sample training set and fine-tuning pre-trained models, and an ensemble classifier for military equipment is built. Experimental results have confirmed that transfer learning saves training time on small samples tasks, resolves issues of model over fitting and strong dependence on data labels simultaneously, and can effectively improve the accuracy of small sample image classification of military equipment based on deep learning.
文章引用：鲁磊纪, 余红霞, 肖红菊, 鲍蕾. 基于迁移学习的军事少样本集成分类研究[J]. 计算机科学与应用, 2024, 14(8): 230-235. https://doi.org/10.12677/csa.2024.148180

参考文献

[1] Youk, G., Oh, J. and Kim, M. (2024) FMA-Net: Flow-Guided Dynamic Filtering and Iterative Feature Refinement with Multi-Attention for Joint Video Super-Resolution and Deblurring. Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle WA, USA, 2024.
https://arxiv.org/abs/2401.03707
[2] Lu, Y.X., Ai, Y., Sheng, Z.Y., et al. (2024) Multi-Stage Speech Bandwidth Extension with Flexible Sampling Rates Control.
https://arxiv.org/abs/2406.02250
[3] 张焕. 基于图像处理与深度学习的典型军事目标识别[D]: [硕士学位论文]. 南京: 南京理工大学, 2021.
[4] 陶志文. 基于深度学习的多战场环境军事人员图像语义分割技术研究[D]: [硕士学位论文]. 北京: 军事科学院, 2021.
[5] Eustratiadis, P., Dudziak, Ł., Li, D. and Hospedales, T. (2024) Neural Fine-Tuning Search for Few-Shot Learning. Proceedings of Conference on Learning Representations, Vienna Austria, 2024.
https://openreview.net/forum?id=T7YV5UZKBc
[6] Wen, H.F., Xing, H. and Simeone, O. (2024) Pre-Training and Personalized Fine-Tuning via Over-the-Air Federated Meta-Learning: Convergence-Generalization Trade-Offs.
https://arxiv.org/abs/2406.11569
[7] Lee, H., Guntara, T.W., Lee, J., et al. (2024) Kernel Metric Learning for In-Sample Off-Policy Evaluation of Deterministic RL Policies. Proceedings of International Conference on Learning Representations, Vienna Austria, 2024.
https://arxiv.org/pdf/2405.18792
[8] Hu, J.S., Jiang, Y.P. and Weng. P. (2024) Revisiting Data Augmentation in Deep Reinforcement Learning. Proceedings of International Conference on Learning Representations, Vienna, Austria, 2024.
https://arxiv.org/pdf/2402.12181v1
[9] Jiang, J.G., Shu, Y., Wang, J.M., et al. (2022) Transferability in Deep Learning: A Survey.
https://arxiv.org/abs/2201.05867
[10] Liu, X., Liu, Z., Wang, G., Cai, Z. and Zhang, H. (2018) Ensemble Transfer Learning Algorithm. IEEE Access, 6, 2389-2396. [Google Scholar] [CrossRef] 
[11] Glorot, X. and Bengio, Y. (2010) Understanding the Difficulty of Training Deep Feedforward Neural Networks. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, 13-15 May 2010, Sardinia, Italy, 249-256.