# Efficient data augmentation using graph imputation neural networks

## TL;DR

This paper introduces an efficient semi-supervised data augmentation method using graph imputation neural networks (GINN), which reconstructs damaged data points by leveraging dataset similarities, significantly improving performance on benchmark datasets.

## Contribution

The paper presents a novel data augmentation technique based on GINN that effectively reconstructs damaged data points by exploiting graph-based similarities, enabling up to 10x dataset augmentation.

## Key findings

- Significant performance improvements over fully-supervised models.
- Ability to augment datasets up to 10 times.
- Effective reconstruction of damaged data points using GINN.

## Abstract

Recently, data augmentation in the semi-supervised regime, where unlabeled data vastly outnumbers labeled data, has received a considerable attention. In this paper, we describe an efficient technique for this task, exploiting a recent framework we proposed for missing data imputation called graph imputation neural network (GINN). The key idea is to leverage both supervised and unsupervised data to build a graph of similarities between points in the dataset. Then, we augment the dataset by severely damaging a few of the nodes (up to 80\% of their features), and reconstructing them using a variation of GINN. On several benchmark datasets, we show that our method can obtain significant improvements compared to a fully-supervised model, and we are able to augment the datasets up to a factor of 10x. This points to the power of graph-based neural networks to represent structural affinities in the samples for tasks of data reconstruction and augmentation.