# Measuring the Data Efficiency of Deep Learning Methods

## TL;DR

This paper introduces a new protocol to benchmark data efficiency of deep learning algorithms, revealing scenarios where hierarchical graph-based methods outperform CNNs in low-data regimes.

## Contribution

It proposes a novel experimental protocol for measuring data efficiency and compares CNNs with HiGSFA on MNIST and Omniglot datasets.

## Key findings

- HiGSFA outperforms CNNs with very small training sets on MNIST.
- CNNs perform better with larger training sets.
- Results suggest local bottom-up learning can match or surpass gradient-based methods.

## Abstract

In this paper, we propose a new experimental protocol and use it to benchmark the data efficiency --- performance as a function of training set size --- of two deep learning algorithms, convolutional neural networks (CNNs) and hierarchical information-preserving graph-based slow feature analysis (HiGSFA), for tasks in classification and transfer learning scenarios. The algorithms are trained on different-sized subsets of the MNIST and Omniglot data sets. HiGSFA outperforms standard CNN networks when the models are trained on 50 and 200 samples per class for MNIST classification. In other cases, the CNNs perform better. The results suggest that there are cases where greedy, locally optimal bottom-up learning is equally or more powerful than global gradient-based learning.