[14] Been Kim - Interactive and Interpretable Machine Learning Models
The Thesis Review
00:00
The Probabilistic Method Can Avoid Unreliability Issues
Going back to the thesis, in the Bayesian model you had here, like you mentioned giving pixels as what we're giving as the interpretability unit. Here in the thesis, you were using the different features and then the different clusters themselves. Are those still used in certain methods today? Or like, do you think that something about this probabilistic formulation would prevent some of those unreliability issues? Yes and no. I mean, all the methods that we exam in our work is sort of, is proven to be working in with different goals in mind.
Play episode from 30:48
Transcript
