Abstract
A recent study has proposed the "Master Key Filters Hypothesis" for convolutional neural network filters. This paper extends this hypothesis by radically constraining its scope to a single set of just 8 universal filters that depthwise separable convolutional networks inherently converge to. While conventional DS-CNNs employ thousands of distinct trained filters, our analysis reveals these filters are predom- inantly linear shifts (ax+b) of our discovered universal set. Through systematic unsupervised search, we extracted these fundamental patterns across different architectures and datasets. Remarkably, networks initialized with these 8 unique frozen filters achieve over 80% ImageNet accuracy, and even outperform models with thousands of trainable parameters when applied to smaller datasets. The identified master key filters closely match Difference of Gaussians (DoGs), Gaussians, and their derivatives, structures that are not only fundamental to classical image processing but also strikingly similar to receptive fields in mammalian visual systems. Our findings provide compelling evidence that depthwise convolutional layers naturally gravitate toward this fundamental set of spatial operators regardless of task or architecture. This work offers new insights for understanding generalization and transfer learning through the universal language of these master key filters.
Abstract (translated)
最近的一项研究提出了卷积神经网络滤波器的“主密钥滤波假设”。本文在此基础上进行了扩展，通过激进地将该假设的范围限制为深度可分离卷积网络内在收敛到的一组仅有的8个通用滤波器来实现。尽管传统的DS-CNNs 使用了数千种不同的训练滤波器，但我们的分析表明，这些滤波器主要是我们发现的通用集合中线性变换（ax+b）的结果。通过系统的无监督搜索，我们在不同架构和数据集上提取了这些基本模式。值得注意的是，使用这8个独特的冻结滤波器初始化的网络在ImageNet数据集上的准确率超过了80%，甚至在应用于较小的数据集时也优于具有数千可训练参数的模型。 我们识别出的主密钥滤波器与高斯差（DoGs）、高斯函数及其导数非常接近，这些结构不仅是经典图像处理的基本组成部分，而且与哺乳动物视觉系统中的感受野惊人地相似。我们的发现提供了强有力的证据表明，深度卷积层无论在何种任务或架构下都会自然趋向于这一基本的空间操作集。 这项工作通过主密钥滤波器的通用语言为理解泛化和迁移学习提供了新的见解。
URL
https://arxiv.org/abs/2509.11711
