Been Kim: Interpretable Machine Learning
The Gradient: Perspectives on AI
00:00
Inherently Interpretable Models Are One Way Communication
In my thesis, I made inherently interpretable models based on the prototypes. And then with that, I realized that the interpretable models are one way communication. Models tell humans, okay, this is what I think. Now there's no way for humans to tell back to the models, wait a second, that's not what I think and please incorporate. There's no way to do that in their models. So my final chapter in my thesis is interactive machine learning, which I call IBCM,. Where we enable this pipeline, pretty simple pipeline for humans to communicate back to machine.
Transcript
Play full episode
