Optimizing Kernel-Target Alignment for cloud detection in multispectral satellite images

TL;DR

This paper investigates the optimization landscape of Kernel-Target Alignment in quantum classifiers, using multispectral satellite data for cloud detection, revealing how data quantity influences the extremum's width.

Contribution

It introduces a simple model to analyze the optimization landscape of Kernel-Target Alignment and studies its behavior in the context of satellite image cloud detection.

Findings

Underparameterized circuits have many local extrema or flat landscapes.

The width of the global extremum peak depends on the amount of data.

Experimental results on satellite data validate the theoretical insights.

Abstract

The optimization of Kernel-Target Alignment (TA) has been recently proposed as a way to reduce the number of hardware resources in quantum classifiers. It allows to exchange highly expressive and costly circuits to moderate size, task oriented ones. In this work we propose a simple toy model to study the optimization landscape of the Kernel-Target Alignment. We find that for underparameterized circuits the optimization landscape possess either many local extrema or becomes flat with narrow global extremum. We find the dependence of the width of the global extremum peak on the amount of data introduced to the model. The experimental study was performed using multispectral satellite data, and we targeted the cloud detection task, being one of the most fundamental and important image analysis tasks in remote sensing.