STAT689-Project

Handling mislabeled training data for classification

Team Members:

• Siddharth Subramaniyam
• Shubham Jain

Report

Report is present in Report.ipynb file

Code

1. General Approach to Identify Mislabeling:

   • Code is in script.py, main.py, util.py and dataset.py
   • Dataset for this is present in mnist-data folder

2. Rankprunning Approach

   • Rankpruning implementation - rankpruning.py
   • Comparison and analysis - rp_comparison.py
   • Generic classifier interface for use with RP - classifier_for_rp.py