An Empirical Investigation of Model-to-Model Distribution Shifts in Trained Convolutional Filters

TL;DR

This paper empirically investigates distribution shifts in trained convolutional filters across various models and datasets, providing a large dataset and insights into how these shifts relate to model generalization and transfer learning.

Contribution

It introduces a large dataset of trained CNN filters and analyzes distribution shifts in these filters across different axes, offering new insights into model robustness and transfer learning.

Findings

Distribution shifts vary with data type, task, architecture, and layer depth.

Some filters show significant distribution shifts, others remain stable.

The properties of filter distributions can inform robustness and transfer learning strategies.

Abstract

We present first empirical results from our ongoing investigation of distribution shifts in image data used for various computer vision tasks. Instead of analyzing the original training and test data, we propose to study shifts in the learned weights of trained models. In this work, we focus on the properties of the distributions of dominantly used 3x3 convolution filter kernels. We collected and publicly provide a data set with over half a billion filters from hundreds of trained CNNs, using a wide range of data sets, architectures, and vision tasks. Our analysis shows interesting distribution shifts (or the lack thereof) between trained filters along different axes of meta-parameters, like data type, task, architecture, or layer depth. We argue, that the observed properties are a valuable source for further investigation into a better understanding of the impact of shifts in the input data to the generalization abilities of CNN models and novel methods for more robust transfer-learning in this domain. Data available at: https://github.com/paulgavrikov/CNN-Filter-DB/.