CaltechAUTHORS
  A Caltech Library Service

Out-of-Distribution Detection Using Neural Rendering Generative Models

Huang, Yujia and Dai, Sihui and Nguyen, Tan and Baraniuk, Richard G. and Anandkumar, Anima (2019) Out-of-Distribution Detection Using Neural Rendering Generative Models. . (Unpublished) https://resolver.caltech.edu/CaltechAUTHORS:20190905-154251302

[img] PDF - Submitted Version
See Usage Policy.

6Mb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20190905-154251302

Abstract

Out-of-distribution (OoD) detection is a natural downstream task for deep generative models, due to their ability to learn the input probability distribution. There are mainly two classes of approaches for OoD detection using deep generative models, viz., based on likelihood measure and the reconstruction loss. However, both approaches are unable to carry out OoD detection effectively, especially when the OoD samples have smaller variance than the training samples. For instance, both flow based and VAE models assign higher likelihood to images from SVHN when trained on CIFAR-10 images. We use a recently proposed generative model known as neural rendering model (NRM) and derive metrics for OoD. We show that NRM unifies both approaches since it provides a likelihood estimate and also carries out reconstruction in each layer of the neural network. Among various measures, we found the joint likelihood of latent variables to be the most effective one for OoD detection. Our results show that when trained on CIFAR-10, lower likelihood (of latent variables) is assigned to SVHN images. Additionally, we show that this metric is consistent across other OoD datasets. To the best of our knowledge, this is the first work to show consistently lower likelihood for OoD data with smaller variance with deep generative models.


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/1907.04572arXivDiscussion Paper
ORCID:
AuthorORCID
Baraniuk, Richard G.0000-0002-0721-8999
Additional Information:A. Anandkumar is supported in part by Bren endowed chair, Darpa PAI, Raytheon, and Microsoft, Google and Adobe faculty fellowships.
Funders:
Funding AgencyGrant Number
Bren ProfessorshipUNSPECIFIED
Defense Advanced Research Projects Agency (DARPA)UNSPECIFIED
Raytheon CompanyUNSPECIFIED
Microsoft Faculty FellowshipUNSPECIFIED
Google Faculty Research AwardUNSPECIFIED
AdobeUNSPECIFIED
Record Number:CaltechAUTHORS:20190905-154251302
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20190905-154251302
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:98454
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:05 Sep 2019 23:18
Last Modified:03 Oct 2019 21:41

Repository Staff Only: item control page