Bridging the Gap: Simultaneous Fine Tuning for Data Re-Balancing

TL;DR

This paper introduces a simultaneous fine-tuning approach that uses supplemental data to address class imbalance in real-world classification tasks, improving neural network training when data is limited or diverse.

Contribution

The paper proposes a novel fine-tuning strategy leveraging supplemental data to better balance classes without discarding diverse large datasets.

Findings

Improved classification accuracy on synthetic aperture sonar data.

Effective handling of data imbalance with supplemental data.

Outperforms traditional under-sampling methods.

Abstract

There are many real-world classification problems wherein the issue of data imbalance (the case when a data set contains substantially more samples for one/many classes than the rest) is unavoidable. While under-sampling the problematic classes is a common solution, this is not a compelling option when the large data class is itself diverse and/or the limited data class is especially small. We suggest a strategy based on recent work concerning limited data problems which utilizes a supplemental set of images with similar properties to the limited data class to aid in the training of a neural network. We show results for our model against other typical methods on a real-world synthetic aperture sonar data set. Code can be found at github.com/JohnMcKay/dataImbalance.