# On Emergence of Clean-Priority Learning in Early Stopped Neural Networks

arXiv: https://arxiv.org/abs/2306.02533 · HTML: https://mlsanity.com/paper/2306.02533/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 20 references that are in the map.
- First posted on arXiv 5 Jun 2023.

## Builds on (most cited first, 11 of 11)

- [A Closer Look at Memorization in Deep Networks](https://mlsanity.com/paper/1706.05394/) (2017)
- [Speech Recognition with Deep Recurrent Neural Networks](https://mlsanity.com/paper/1303.5778/) (2013)
- [Loss landscapes and optimization in over-parameterized non-linear systems and neural networks](https://mlsanity.com/paper/2003.00307/) (2020)
- [To understand deep learning we need to understand kernel learning](https://mlsanity.com/paper/1802.01396/) (2018)
- [A Theoretically Grounded Application of Dropout in Recurrent Neural Networks](https://mlsanity.com/paper/1512.05287/) (2015)
- [Gradient Descent with Early Stopping is Provably Robust to Label Noise for Overparameterized Neural Networks](https://mlsanity.com/paper/1903.11680/) (2019)
- [On the linearity of large non-linear models: when and why the tangent kernel is constant](https://mlsanity.com/paper/2010.01092/) (2020)
- [Early-stopped neural networks are consistent](https://mlsanity.com/paper/2106.05932/) (2021)
- [Transition to Linearity of Wide Neural Networks is an Emerging Property of Assembling Weak Models](https://mlsanity.com/paper/2203.05104/) (2022)
- [On Optimal Early Stopping: Over-informative versus Under-informative Parametrization](https://mlsanity.com/paper/2202.09885/) (2022)
- [Transition to Linearity of General Neural Networks with Directed Acyclic Graph Architecture](https://mlsanity.com/paper/2205.11786/) (2022)

## Built on by (most cited first, 3 of 3)

- [Why is SAM Robust to Label Noise?](https://mlsanity.com/paper/2405.03676/) (2024)
- [On the Role of Label Noise in the Feature Learning Process](https://mlsanity.com/paper/2505.18909/) (2025)
- [Optimization dependent generalization bound for ReLU networks based on sensitivity in the tangent bundle](https://mlsanity.com/paper/2310.17378/) (2023)

## Abstract (arXiv)

When random label noise is added to a training dataset, the prediction error of a neural network on a label-noise-free test dataset initially improves during early training but eventually deteriorates, following a U-shaped dependence on training time. This behaviour is believed to be a result of neural networks learning the pattern of clean data first and fitting the noise later in the training, a phenomenon that we refer to as clean-priority learning. In this study, we aim to explore the learning dynamics underlying this phenomenon. We theoretically demonstrate that, in the early stage of training, the update direction of gradient descent is determined by the clean subset of training data, leaving the noisy subset has minimal to no impact, resulting in a prioritization of clean learning. Moreover, we show both theoretically and experimentally, as the clean-priority learning goes on, the dominance of the gradients of clean samples over those of noisy samples diminishes, and finally results in a termination of the clean-priority learning and fitting of the noisy samples.