Timezone: »

 
Poster
A Benchmark for Interpretability Methods in Deep Neural Networks
Sara Hooker · Dumitru Erhan · Pieter-Jan Kindermans · Been Kim

Tue Dec 10 10:45 AM -- 12:45 PM (PST) @ East Exhibition Hall B + C #159
in Deep Learning -- Visualization or Exposition Techniques for Deep Networks »

We propose an empirical measure of the approximate accuracy of feature importance estimates in deep neural networks. Our results across several large-scale image classification datasets show that many popular interpretability methods produce estimates of feature importance that are not better than a random designation of feature importance. Only certain ensemble based approaches---VarGrad and SmoothGrad-Squared---outperform such a random assignment of importance. The manner of ensembling remains critical, we show that some approaches do no better then the underlying method but carry a far higher computational burden.

Keywords: Deep Learning -> Visualization or Exposition Techniques for Deep Networks · Data, Challenges, Implementations, and Software -> Benchmarks

Author Information

Sara Hooker (Google Brain)

I am a Research Scholar at Google doing machine learning research. My research interests include algorithm transparency, security and privacy.

Dumitru Erhan (Google Brain)
Pieter-Jan Kindermans (Google Brain)
Been Kim (Google)

More from the Same Authors