How much data do you need? Part 2: Predicting DL class specific training dataset sizes

TL;DR

This paper introduces a method to predict the performance of classification models based on the distribution of training examples across classes, using models like powerlaw curves and applying it to datasets like CIFAR10 and EMNIST.

Contribution

It proposes a novel algorithm for estimating class-specific dataset sizes needed for optimal model performance, extending traditional models with a new combinatorial approach.

Findings

Effective prediction of class-specific dataset sizes.

Application to CIFAR10 and EMNIST datasets.

Model fits well with experimental data.

Abstract

This paper targets the question of predicting machine learning classification model performance, when taking into account the number of training examples per class and not just the overall number of training examples. This leads to the a combinatorial question, which combinations of number of training examples per class should be considered, given a fixed overall training dataset size. In order to solve this question, an algorithm is suggested which is motivated from special cases of space filling design of experiments. The resulting data are modeled using models like powerlaw curves and similar models, extended like generalized linear models i.e. by replacing the overall training dataset size by a parametrized linear combination of the number of training examples per label class. The proposed algorithm has been applied on the CIFAR10 and the EMNIST datasets.