Robust Loss Functions for Training Decision Trees with Noisy Labels

Jonathan Wilton, Nan Ye

We consider training decision trees using noisily labeled data, focusing on loss functions that can lead to robust learning algorithms. Our contributions are threefold. First, we o…