# Ghost Loss to Question the Reliability of Training Data

arXiv: https://arxiv.org/abs/2109.01504 · HTML: https://mlsanity.com/paper/2109.01504/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 22 references that are in the map.
- First posted on arXiv 3 Sep 2021.

## Builds on (most cited first, 10 of 10)

- [Densely Connected Convolutional Networks](https://mlsanity.com/paper/1608.06993/) (2016)
- [Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms](https://mlsanity.com/paper/1708.07747/) (2017)
- [Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels](https://mlsanity.com/paper/1805.07836/) (2018)
- [Dynamic Routing Between Capsules](https://mlsanity.com/paper/1710.09829/) (2017)
- [Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach](https://mlsanity.com/paper/1609.03683/) (2016)
- [CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise](https://mlsanity.com/paper/1711.07131/) (2017)
- [Deep learning with noisy labels: exploring techniques and remedies in medical image analysis](https://mlsanity.com/paper/1912.02911/) (2019)
- [Combating Label Noise in Deep Learning Using Abstention](https://mlsanity.com/paper/1905.10964/) (2019)
- [HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules](https://mlsanity.com/paper/1806.06519/) (2018)
- [DropMax: Adaptive Variational Softmax](https://mlsanity.com/paper/1712.07834/) (2017)

## Abstract (arXiv)

Supervised image classification problems rely on training data assumed to have been correctly annotated; this assumption underpins most works in the field of deep learning. In consequence, during its training, a network is forced to match the label provided by the annotator and is not given the flexibility to choose an alternative to inconsistencies that it might be able to detect. Therefore, erroneously labeled training images may end up ``correctly'' classified in classes which they do not actually belong to. This may reduce the performances of the network and thus incite to build more complex networks without even checking the quality of the training data. In this work, we question the reliability of the annotated datasets. For that purpose, we introduce the notion of ghost loss, which can be seen as a regular loss that is zeroed out for some predicted values in a deterministic way and that allows the network to choose an alternative to the given label without being penalized. After a proof of concept experiment, we use the ghost loss principle to detect confusing images and erroneously labeled images in well-known training datasets (MNIST, Fashion-MNIST, SVHN, CIFAR10) and we provide a new tool, called sanity matrix, for summarizing these confusions.