On Training Sample Memorization: Lessons from Benchmarking Generative Modeling with a Large-scale Competition
| العنوان: | On Training Sample Memorization: Lessons from Benchmarking Generative Modeling with a Large-scale Competition |
|---|---|
| المؤلفون: | Hsuan-Tien Lin, Colin Raffel, Wendy Chih-wen Kan, Ching-Yuan Bai |
| المصدر: | Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. |
| بيانات النشر: | ACM, 2021. |
| سنة النشر: | 2021 |
| مصطلحات موضوعية: | FOS: Computer and information sciences, Computer Science - Machine Learning, Computer science, business.industry, Scale (chemistry), media_common.quotation_subject, Sample (statistics), Benchmarking, Machine learning, computer.software_genre, Memorization, Machine Learning (cs.LG), Metric (mathematics), Benchmark (computing), Quality (business), Artificial intelligence, business, computer, Generative grammar, media_common |
| الوصف: | Many recent developments on generative models for natural images have relied on heuristically-motivated metrics that can be easily gamed by memorizing a small sample from the true distribution or training a model directly to improve the metric. In this work, we critically evaluate the gameability of these metrics by designing and deploying a generative modeling competition. Our competition received over 11000 submitted models. The competitiveness between participants allowed us to investigate both intentional and unintentional memorization in generative modeling. To detect intentional memorization, we propose the ``Memorization-Informed Fr\'echet Inception Distance'' (MiFID) as a new memorization-aware metric and design benchmark procedures to ensure that winning submissions made genuine improvements in perceptual quality. Furthermore, we manually inspect the code for the 1000 top-performing models to understand and label different forms of memorization. Our analysis reveals that unintentional memorization is a serious and common issue in popular generative models. The generated images and our memorization labels of those models as well as code to compute MiFID are released to facilitate future studies on benchmarking generative models. Comment: In Proceedings of the 27th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), August 2021 |
| DOI: | 10.1145/3447548.3467198 |
| URL الوصول: | https://explore.openaire.eu/search/publication?articleId=doi_dedup___::2d0be9ecb30086801d2f5853a67021c1 https://doi.org/10.1145/3447548.3467198 |
| Rights: | OPEN |
| رقم الانضمام: | edsair.doi.dedup.....2d0be9ecb30086801d2f5853a67021c1 |
| قاعدة البيانات: | OpenAIRE |
| DOI: | 10.1145/3447548.3467198 |
|---|