GRIFT: A graphical model for inferring visual classification features from human data [Poster ID W50]
Michael G. Ross and Andrew L. Cohen How do people classify images? targets class 1 University of Massachusetts Amherst GRIFT recovers non-linear classification strategies from human data
S i Fi i N 0

noisy sample

features recovered from human subject

class 2

is this an example of class 1 or 2? weight vector

i

Linearly?

·
Non-linearly? feature detectors

> 0?

C

Visit our poster for details and results on other image classes...

class 1 or 2?