条条大路通罗马?探究生成式图像模型潜在空间之间的表征相似性
All Roads Lead to Rome? Exploring Representational Similarities Between Latent Spaces of Generative Image Models
AI总结:
本文通过训练线性映射拼接四种生成模型的潜在空间,发现性能良好的模型间线性映射能保留视觉信息,性别属性表征最相似,且NF的潜在表征在训练早期收敛。
摘要:
不同的生成式图像模型是否在暗中学习相似的基础表征?我们通过测量四种不同模型(VAE、GAN、归一化流(NFs)和扩散模型(DMs))的潜在空间相似性来研究这一问题。我们的方法包括在冻结的潜在空间之间训练线性映射,以“拼接”任意编码器-解码器对,并在由此产生的“拼接”模型上测量基于输出和基于探针的指标。我们的主要发现是:性能良好的模型的潜在空间之间的线性映射能保留大部分视觉信息,即使潜在尺寸不同;对于CelebA模型,性别是最相似的可探针属性。最后,我们在一个NF上表明,潜在空间表征在训练早期就趋于收敛。
英文摘要:
Do different generative image models secretly learn similar underlying representations? We investigate this by measuring the latent space similarity of four different models: VAEs, GANs, Normalizing Flows (NFs), and Diffusion Models (DMs). Our methodology involves training linear maps between frozen latent spaces to "stitch" arbitrary pairs of encoders and decoders and measuring output-based and probe-based metrics on the resulting "stitched'' models. Our main findings are that linear maps between latent spaces of performant models preserve most visual information even when latent sizes differ; for CelebA models, gender is the most similarly represented probe-able attribute. Finally we show on an NF that latent space representations converge early in training.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。