SolidMark: Evaluating Image Memorization in Generative Models
This addresses the need for better evaluation tools in generative AI research to understand and mitigate memorization issues, though it is incremental as it builds on prior work on memorization metrics.
The paper tackles the problem of evaluating image memorization in diffusion models, which suffer from biased metrics and poor detection of specific memorized images, by introducing SolidMark, a novel method that provides per-image memorization scores and re-evaluates existing mitigation techniques, showing it can assess fine-grained pixel-level memorization.
Recent works have shown that diffusion models are able to memorize training images and emit them at generation time. However, the metrics used to evaluate memorization and its mitigation techniques suffer from dataset-dependent biases and struggle to detect whether a given specific image has been memorized or not. This paper begins with a comprehensive exploration of issues surrounding memorization metrics in diffusion models. Then, to mitigate these issues, we introduce $\rm \style{font-variant: small-caps}{SolidMark}$, a novel evaluation method that provides a per-image memorization score. We then re-evaluate existing memorization mitigation techniques. We also show that $\rm \style{font-variant: small-caps}{SolidMark}$ is capable of evaluating fine-grained pixel-level memorization. Finally, we release a variety of models based on $\rm \style{font-variant: small-caps}{SolidMark}$ to facilitate further research for understanding memorization phenomena in generative models. All of our code is available at https://github.com/NickyDCFP/SolidMark.