/MLNT

Meta-Learning based Noise-Tolerant Training

Primary LanguagePython

Meta-Learning based Noise-Tolerant Training

This is the PyTorch code for CVPR 2019 Paper:
Learning to Learn from Noisy Labeled Data [pdf]
Junnan Li, Y. Wong, Qi Zhao, M. Kankanhalli
To use the code, please cite our paper.

To run the code, please first download Clothing1M into ./data folder.
Run baseline.py to get the result and model trained with conventional cross entropy loss and SGD.
Run main.py to get the result and model trained with MLNT.