Training Convolutional Networks with Noisy Labels

By Sainbayar Sukhbaatar, Joan Bruna, Manohar Paluri, Lubomir D. Bourdev, Rob Fergus. 2014

674 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Optimization & Training

Citations in the corpus

Temporal Ensembling for Semi-Supervised Learning (2016, background): “For recent work in this area, see, e.g., Sukhbaatar et al. (2014) and Patrini et al. (2016).”
Seeing through the Human Reporting Bias: Visual Classifiers from Noisy Human-Centric Labels (2015, background): “Here we demonstrate that our model is indeed improved by conditioning the noise (relevance) distribution on the input image, in contrast to previous work [36, 47] that estimates noise parameters without conditioning on the image.”
Using Pre-Training Can Improve Model Robustness and Uncertainty (2019, methodology): “We bring clarity to the doubts raised about pre-training by showing that pre-training can improve model robustness to label corruption (Sukhbaatar et al., 2014), class imbalance (Japkowicz, 2000), and adversarial attacks (Szegedy et al.”
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty (2019, methodology): “This is a two-stage loss correction method based on Sukhbaatar et al. [2014] and Patrini et al. [2017].”
Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (2019, background): “While novel changes to architectures have been proposed [57, 61] to combat label noise, our experiments demonstrate that large-scale training of an existing video architecture [14] makes it noise-resilient.”
Learning from Noisy Large-Scale Datasets with Minimal Supervision (2017, background): “[27] both model noise that is conditionally independent from the input image.”
Toward an Integration of Deep Learning and Neuroscience (2016, background): “The crude detection of human faces used together with statistical learning should be analogous to semi-supervised learning (Sukhbaatar et al., 2014) and could allow identifying faces with high certainty.”
MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels (2017, background): “Sukhbaatar et al. (2014) proposed a noise transformation to estimate the noise distribution.”
Direct Uncertainty Prediction for Medical Second Opinions (2018, background): “As this expert disagreement arises from human judgement and bias, the instance x i itself contains features that may give rise to an intrinsic level of disagreement in human evaluation, separate from uniform noise [14], or noise dependent on solely ground truth labels (diagnoses) [16].”
CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise (2017, methodology): “Some methods were developed for directly learning neural network with label noise [1, 3, 14, 20, 22, 27, 32, 35, 41].”
Learning From Noisy Singly-labeled Data (2017, background): “Sukhbaatar et al. (2014); Jindal et al. (2016) learns noise rate as parameters of the model.”
Deep Self-Learning From Noisy Labels (2019, methodology): “in [32] added an extra linear layer to model the transition relationships between true and corrupted labels.”
Learning from data in the mixed adversarial non-adversarial case: Finding the helpers and ignoring the trolls (2022, background): “A recent review of this ﬁeld (Song et al., 2020) characterizes proposed solutions into four categories: robust architectures (e.g., (Sukhbaatar et al., 2014; Xiao et al., 2015)), robust loss functions (e.g., (Ghosh et al., 2017; Liu and Guo, 2020)), robust regularization (e.g., (Jenni and Favaro,…”