Code repository for the robust active label correction paper.
-
Updated
Apr 12, 2018 - Terra
Code repository for the robust active label correction paper.
Tensorflow source code for "CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise" (CVPR 2018)
Contains my experiments for the Game of Deep Learning Hackathon conducted by Analytics Vidhya
Mean Absolute Error Does Not Treat Examples Equally and Gradient Magnitude’s Variance Matters
Implementations of different loss-correction techniques to help deep models learn under class-conditional label noise.
In the context of Deep Learning: What is the right way to conduct example weighting? How do you understand loss functions and so-called theorems on them?
This is the source code for Butterfly: One-step Approach towards Wildly Unsupervised Domain Adaptation (NeurIPS'19 Workshop).
Analysis of robust classification algorithms for overcoming class-dependant labelling noise: Forward, Importance Reweighting and T-revision. We demonstrate methods for estimating the transition matrix in order to obtain better classifier performance when working with noisy data.
AAAI 2021: Robustness of Accuracy Metric and its Inspirations in Learning with Noisy Labels
AAAI 2021: Beyond Class-Conditional Assumption: A Primary Attempt to Combat Instance-Dependent Label Noise
Extra bits of unsanitized code for plotting, training, etc. related to our CVPR 2021 paper "Augmentation Strategies for Learning with Noisy Labels".
[Re] Can gradient clipping mitigate label noise? (ML Reproducibility Challenge 2020)
SREA: Self-Re-Labeling with Embedding Analysis
[TPAMI2022 & NeurIPS2020] Official implementation of Self-Adaptive Training
[ACM MM 2021 Oral Presentation] A unified framework for co-training-based noisy label learning methods.
[CVPR 2021] Code for "Augmentation Strategies for Learning with Noisy Labels".
PyTorch Implementation of Robust Cross Entropy Loss (Loss Correction for Label Noise)
Add a description, image, and links to the label-noise topic page so that developers can more easily learn about it.
To associate your repository with the label-noise topic, visit your repo's landing page and select "manage topics."