Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: A Note on the Inception Score
(Submitted on 6 Jan 2018 (v1), last revised 21 Jun 2018 (this version, v2))
Abstract: Deep generative models are powerful tools that have produced impressive results in recent years. These advances have been for the most part empirically driven, making it essential that we use high quality evaluation metrics. In this paper, we provide new insights into the Inception Score, a recently proposed and widely used evaluation metric for generative models, and demonstrate that it fails to provide useful guidance when comparing models. We discuss both suboptimalities of the metric itself and issues with its application. Finally, we call for researchers to be more systematic and careful when evaluating and comparing generative models, as the advancement of the field depends upon it.
Submission history
From: Shane Barratt [view email][v1] Sat, 6 Jan 2018 05:44:29 GMT (4275kb,D)
[v2] Thu, 21 Jun 2018 14:54:33 GMT (4305kb,D)
Link back to: arXiv, form interface, contact.