Skip to content
The methodOpen data
Back to the collection

Deep Learning is Robust to Massive Label Noise

David Rolnick, Andreas Veit, Serge Belongie, Nir Shavit

2017 · arXiv

Machine learningPsychologyComputer vision
Read the paper
HN score
0
Mentions
3
Readers
3
Months
3
What’s in a score?

From the abstract

Deep neural networks trained on large supervised datasets have led to impressive results in image classification and other tasks. However, well-annotated datasets can be time-consuming and expensive to collect, lending increased interest to larger but noisy datasets that are more easily obtained. In this paper, we show that deep neural networks are capable of generalizing from training data for which true labels are massively outnumbered by incorrect labels. We demonstrate remarkably high test performance after training on corrupted data from MNIST, CIFAR, and ImageNet. For example, on MNIST we obtain test accuracy above 90 percent even after each clean training example has been diluted with 100 randomly-labeled examples. Such behavior holds across multiple patterns of label noise, even when erroneous labels are biased towards confusing classes. We show that training in this regime requires a significant but manageable increase in dataset size that is related to the factor by which correct labels have been diluted. Finally, we provide an analysis of our results that shows how increasing noise decreases the effective batch size.

Metadata: arXiv. Abstracts retain their original authors’ rights.

In the conversation

Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels

3 selected excerpts · 3 total mentions

dougabug
Neutral mention · rule
You need to be very careful about making sweeping generalizations based on a single personal anecdote. The really large data sets typically have very high error rates and sample biases. For instance, Google’s JTF300M is far noisier than ImageNet, which itself is hardly free of errors and biases. Any data set with hundreds of millions to billions of images will generally contain a large proportion of images and labels scraped from the web, w/ automatic filtering or pseudolabeling, perhaps w/ some degree of sampled verification by human labelers. In fact, generally DL is quite tolerant to label noise, especially using modern training methods such as SSL pretraining. https://arxiv.org/pdf/1705.10694.pdf https://proceedings.neurips.cc/paper/2018/file/a19744e268754... https://proceedings.mlr.press/v97/hendrycks19a.html
Original comment Why we chose not to release Stable Diffusion 1.5 as quickly
momenti
Neutral mention · rule
That's not entirely true. Neural networks are fairly robust to noisy training data (a.k.a. garbage).[0] Well, stochastic gradient descent has the noise in its name. More training data can compensate for noisy data to some extent.[1] I'm not sure know if model size can also compensate for noisy data though, but would not be surprised if it did. [0] https://arxiv.org/abs/1705.10694 [1] https://arxiv.org/abs/2202.01994
Original comment Deep learning on electronic medical records is doomed to fail

Keep following the thread.

More in Machine learning

More from the same topics.

2018arXiv

Talent vs Luck: the role of randomness in success and failure

A. Pluchino. A. E. Biondo, A. Rapisarda

The largely dominant meritocratic paradigm of highly competitive Western cultures is rooted on the belief that success is due mainly, if not exclusively, to personal qualities such as talent, intelligence, skills, efforts or risk taking.

Psychology
16 mentions14 months
2.5HN score