Skip to content

wangbobby/noisy_label_learning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

noisy_label_learning

createDataFile.py file produce random point data in setting range. They are clean_100.txt, clean_1000.txt, clean_10000.txt, and clean_100000.txt. We can add testing or taining to the clean data file name, so they can be use for learning optimized parameters.

There are three main files in this folder. They are main_risk.py, main_clean_risk_mesh.py, and main_plot_clean_risk_w_in_diff_rou.py.

main_risk.py find the optimized parameter W for the decision function f hat. And use the optimized function to train the training data and test the testing data. The risks results of training data and testing data in different rou are plotted in a diagram which is shown in final report Figure 2.

main_clean_risk_mesh.py computes all optimized parameter W and those Ws are used to compute the clean data risk in different rou settings. And then, the result is save to clean_risk.txt. Be careful for the input training file name is different to the testing file name. For example, in my setting: clean_file_name = 'clean_1000_training.txt'; and clean_testing_data = readTextFile('clean_1000_testing.txt'; two files must be different.

main_plot_clean_risk_w_in_diff_rou.py plots the file saved by main_clean_risk_mesh.py and the data of the file is plotted in a 3D diagram (Figure three in the final report) which shows that the most area are flat, and this is consistent to the result of lemma one. The file name which save by main_clean_risk_mesh.py must be the same with the file name to be plot in main_plot_clean_risk_w_in_diff_rou.py.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages