https://github.com/Ryo-Ito/Noisy-Labels-Neural-Network

Chainer implementation of Noisy Labels Neural-Network
https://github.com/Ryo-Ito/Noisy-Labels-Neural-Network

Last synced: 2 months ago
JSON representation

Chainer implementation of Noisy Labels Neural-Network

• Host: GitHub
• URL: https://github.com/Ryo-Ito/Noisy-Labels-Neural-Network
• Owner: Ryo-Ito
• Created: 2017-02-28T10:46:09.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-02-28T14:57:07.000Z (over 7 years ago)
• Last Synced: 2024-01-29T09:35:59.066Z (5 months ago)
• Language: Python
• Size: 23.4 KB
• Stars: 5
• Watchers: 3
• Forks: 3
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Lists

• Awesome-Learning-with-Label-Noise - Noisy Label Neural Network2-Chainer
• Awesome-Learning-with-Label-Noise - Noisy Label Neural Network2-Chainer

README

        
# Noisy-Labels-Neural-Network

Chainer implementation of "Training Deep Neural-Networks Based on Unreliable Labels."

Note that this is not an official implementation of the paper.

Training of the network runs by

```bash

$ python train.py

```

The result of relabeling noisy training data would look like below.

![relabel](relabel.png)

# Reference

[1] Bekker, Alan Joseph, and Jacob Goldberger. "Training deep neural-networks based on unreliable labels." Acoustics, Speech and Signal Processing (ICASSP), 2016 IEEE International Conference on. IEEE, 2016.