Skip to content

Implementation of paper: Making Deep Neural Network Robust to Label Noise: a Loss Correction Approach.

Notifications You must be signed in to change notification settings

GarrettLee/label_noise_correction

Repository files navigation

label_noise_correction

Implementation of paper: Making Deep Neural Network Robust to Label Noise: a Loss Correction Approach.

Requirements:

  • Python 2.7
  • TensorFlow 1.4
  • Matplotlb
  • Numpy

Usage

  • Train all models and evaluate all the tests with: python experiment_mnist.py, or with bash script/run_experiment_mnist for faster training and testing. When this is finished, 4 files named backward.npy, backward_t.npy, cross_entropy.npy, forward.npy, forward_t.npy should have been created under the path ./result/mnist/.
  • Show the result with: python show_result_of_mnist_experiment.py.

Result

This is the result of Fully connected network on MNIST. Notice that when N=0.5, the parametric matrix T is singular. results.png

About

Implementation of paper: Making Deep Neural Network Robust to Label Noise: a Loss Correction Approach.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages