HYPE: A Benchmark for Human eYe Perceptual Evaluation of Generative Models
Published in Advances in Neural Information Processing Systems, 2019
This paper introduces a novel crowdsourcing framework to scalably and accurately evaluate human perception of generative ML models. HYPE is a more direct measurement of human judgement than automated proxies, and it is cheaper and more consistent than other human evaluations.
Recommended citation: Sharon Zhou*, Mitchell Gordon*, Ranjay Krishna, Austin Narcomey, Li Fei-Fei, Michael Bernstein. "HYPE: A Benchmark for Human eYe Perceptual Evaluation of Generative Models." Advances in Neural Information Processing Systems. 2019. http://papers.nips.cc/paper/8605-hype-a-benchmark-for-human-eye-perceptual-evaluation-of-generative-models