Mode Collapse

انهيار الأنماط

ظاهرة في تدريب الشبكات التوليدية التنافسية يقتصر فيها الموّلد على إنتاج عدد محدود من المخرجات المتشابهة بدلاً من تغطية التنوّع الكامل لتوزيع البيانات.

A failure mode in GAN training where the Generator produces only a small subset of similar outputs instead of covering the full diversity of the data distribution.

Also translated asانطواء الأنماط، تقلّص التنوّع التوليدي

First appears in this corpus in: Auto-Encoding Variational Bayes (2013)

Appears in these papers

beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework2017 in the sky ✦
Classifier-Free Diffusion Guidance2022 in the sky ✦
Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks2017 in the sky ✦
Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks2015 in the sky ✦
Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks2015 in the sky ✦
Denoising Diffusion Probabilistic Models2020 in the sky ✦
Emerging Properties in Self-Supervised Vision Transformers2021 in the sky ✦
Generative Adversarial Networks2014 in the sky ✦
Generative Adversarial Networks2014 in the sky ✦
mixup: Beyond Empirical Risk Minimization2018 in the sky ✦
Progressive Growing of GANs for Improved Quality, Stability, and Variation2018 in the sky ✦
Progressive Growing of GANs for Improved Quality, Stability, and Variation2018 in the sky ✦
Generative Modeling by Estimating Gradients of the Data Distribution2019 in the sky ✦
Auto-Encoding Variational Bayes2013 in the sky ✦
Wasserstein GAN2017 in the sky ✦
Wasserstein GAN2017 in the sky ✦