Graph convolutional networks for learning with few clean and many noisy labels

TL;DR

This paper introduces a GCN-based method to improve learning from limited clean labels and abundant noisy data by modeling data structure with graphs and weighting examples based on predicted relevance.

Contribution

It proposes a novel GCN approach to identify clean data within noisy labels, enhancing few-shot learning performance with minimal clean data.

Findings

Significant accuracy improvement over uncleaned data

Effective in few-shot learning with noisy labels

GCN-based relevance weighting enhances classifier training

Abstract

In this work we consider the problem of learning a classifier from noisy labels when a few clean labeled examples are given. The structure of clean and noisy data is modeled by a graph per class and Graph Convolutional Networks (GCN) are used to predict class relevance of noisy examples. For each class, the GCN is treated as a binary classifier, which learns to discriminate clean from noisy examples using a weighted binary cross-entropy loss function. The GCN-inferred "clean" probability is then exploited as a relevance measure. Each noisy example is weighted by its relevance when learning a classifier for the end task. We evaluate our method on an extended version of a few-shot learning problem, where the few clean examples of novel classes are supplemented with additional noisy data. Experimental results show that our GCN-based cleaning process significantly improves the classification accuracy over not cleaning the noisy data, as well as standard few-shot classification where only few clean examples are used.