用预定义的边缘线特征滤波器提升CNN泛化能力，小数据下准确率提高5-11%。
Enhancing Generalization in Convolutional Neural Networks through Regularization with Edge and Line Features
- 用固定3×3边缘线滤波器替代可学习卷积核，强制模型关注几何特征。
- 在四个细粒度分类数据集上，测试准确率提升5-11个百分点，参数量不变。
- 滤波器数量≥9时效果最佳，但维度数对性能影响不大，适合小样本场景。
本文提出一种新型正则化方法，引导卷积神经网络（CNN）在隐藏层中优先利用边缘和线性特征。通过将卷积层约束为仅使用预定义的边缘与线检测滤波器，而非学习任意卷积核，实现对模型的有意偏差，从而提升泛化能力，尤其在小数据集上表现显著。实验采用预定义滤波器模块（Pre-defined Filter Modules），以固定的一组3×3边缘和线性滤波器对输入进行卷积，随后通过ReLU消除未激活响应信息，再经1×1卷积生成线性组合。值得注意的是，这些预定义滤波器在整个训练过程中保持不变。结果表明，滤波器集合的维度数对识别性能影响较小，但集合规模至关重要：九个或更多滤波器时达到最优效果。在四个具有挑战性的细粒度分类数据集上，使用相同可训练参数量的情况下，测试准确率平均提升5-11个百分点。
原文摘要 · Abstract (English)
This paper proposes a novel regularization approach to bias Convolutional Neural Networks (CNNs) toward utilizing edge and line features in their hidden layers. Rather than learning arbitrary kernels, we constrain the convolution layers to edge and line detection kernels. This intentional bias regularizes the models, improving generalization performance, especially on small datasets. As a result, test accuracies improve by margins of 5-11 percentage points across four challenging fine-grained classification datasets with limited training data and an identical number of trainable parameters. Instead of traditional convolutional layers, we use Pre-defined Filter Modules, which convolve input data using a fixed set of 3x3 pre-defined edge and line filters. A subsequent ReLU erases information that did not trigger any positive response. Next, a 1x1 convolutional layer generates linear combinations. Notably, the pre-defined filters are a fixed component of the architecture, remaining unchanged during the training phase. Our findings reveal that the number of dimensions spanned by the set of pre-defined filters has a low impact on recognition performance. However, the size of the set of filters matters, with nine or more filters providing optimal results.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。