Understanding Learned Models by Identifying Important Features at the Right Resolution

Kyubin Lee, Akshay Sood, Mark Craven

In many application domains, it is important to characterize how complex learned models make their decisions across the distribution of instances. One way to do this is to identify…