Learning to Aggregate and Refine Noisy Labels for Visual Sentiment Analysis

TL;DR

This paper introduces a robust learning method for visual sentiment analysis that uses an external memory of prototypes to filter noisy labels, improving model generalization despite noisy crowd-sourced data.

Contribution

It proposes a novel memory-based approach to aggregate and filter noisy sentiment labels during training, enhancing deep model robustness in visual sentiment analysis.

Findings

The method effectively filters noisy labels during training.

Prototypes guide the model to prevent overfitting to noise.

Benchmark results demonstrate improved performance with noisy labels.

Abstract

Visual sentiment analysis has received increasing attention in recent years. However, the dataset's quality is a concern because the sentiment labels are crowd-sourcing, subjective, and prone to mistakes, and poses a severe threat to the data-driven models, especially the deep neural networks. The deep models would generalize poorly on the testing cases when trained to over-fit the training samples with noisy sentiment labels. Inspired by the recent progress on learning with noisy labels, we propose a robust learning method to perform robust visual sentiment analysis. Our method relies on external memory to aggregate and filters noisy labels during training. The memory is composed of the prototypes with corresponding labels, which can be updated online. The learned prototypes and their labels can be regarded as denoising features and labels for the local regions and can guide the training process to prevent the model from overfitting the noisy cases. We establish a benchmark for visual sentiment analysis with label noise using publicly available datasets. The experiment results of the proposed benchmark settings comprehensively show the effectiveness of our method.