Label Denoising with Large Ensembles of Heterogeneous Neural Networks

TL;DR

This paper presents a large ensemble approach for video classification on YouTube-8M, utilizing knowledge distillation and mixup to handle noisy labels and improve model performance within hardware constraints.

Contribution

It introduces a novel ensemble-based solution for large-scale video classification that effectively manages noisy labels using knowledge distillation and data augmentation techniques.

Findings

Large ensemble improves classification accuracy

Knowledge distillation effectively handles noisy labels

Mixup enhances model robustness

Abstract

Despite recent advances in computer vision based on various convolutional architectures, video understanding remains an important challenge. In this work, we present and discuss a top solution for the large-scale video classification (labeling) problem introduced as a Kaggle competition based on the YouTube-8M dataset. We show and compare different approaches to preprocessing, data augmentation, model architectures, and model combination. Our final model is based on a large ensemble of video- and frame-level models but fits into rather limiting hardware constraints. We apply an approach based on knowledge distillation to deal with noisy labels in the original dataset and the recently developed mixup technique to improve the basic models.