Analysis of convolutional neural network image classifiers in a hierarchical max-pooling model with additional local pooling

TL;DR

This paper introduces a hierarchical max-pooling model with local pooling for image classification, analyzing its convergence and performance on simulated and real data, offering insights into CNN classifier behavior.

Contribution

The paper proposes a novel hierarchical max-pooling model with local pooling, enhancing the ability to combine image parts with variable distances, and compares various CNN classifiers' convergence rates.

Findings

The hierarchical model effectively combines image parts at variable distances.

Different CNN classifiers show varying convergence rates.

The model performs well on both simulated and real datasets.

Abstract

Image classification is considered, and a hierarchical max-pooling model with additional local pooling is introduced. Here the additional local pooling enables the hierachical model to combine parts of the image which have a variable relative distance towards each other. Various convolutional neural network image classifiers are introduced and compared in view of their rate of convergence. The finite sample size performance of the estimates is analyzed by applying them to simulated and real data.