让生成模型创造既陌生又易学的新概念，提升创意能力。
Seeking the Unfamiliar but Memorable: Conceptual Creativity as Meta-Learning
- 用创作者-评估者框架，通过元学习优化创意产出
- 仅用少量样本就能让评估者快速适应新概念，奖励创作者
- 无需额外语言条件，可生成原模型不会的风格与概念组合
什么是真正的概念创造，而非简单复现熟悉内容？对同一提示反复采样生成模型，通常只会产生风格相似、内容常见的变体。我们提出：创造力体现在生成对适应性观察者而言初看陌生、但经少数样本即可快速掌握的刺激。为此，我们构建了创作者-评估者配对框架：创作者生成候选内容，评估者通过几轮内部学习步骤进行适应，其性能提升作为创作者优化的奖励信号。我们在实验中以扩散模型为创作者，使用在MNIST上的自编码器评估者和带低秩适配器的CLIP评估者处理自然图像。创作者保持冻结，无额外语言条件；仅靠元学习梯度，即可生成原模型未具备的风格变化与概念组合。
原文摘要 · Abstract (English)
What does it mean to create a new concept, rather than retrieve a familiar one? Repeatedly sampling a generative model at the same prompt produces variations with similar styles and typical content. We propose that creativity is the production of stimuli that are unfamiliar to an adaptive observer at first sight, but quickly learnable from a few exposures. We formalize this as a Creator-Appraiser pair: a Creator generates a candidate, an Appraiser adapts to it for a few inner-loop learning steps, and the Appraiser's improvement becomes the reward the Creator optimizes through. We instantiate the framework with diffusion as the Creator, an autoencoder Appraiser on MNIST, and a CLIP Appraiser with a low-rank adapter for natural images. The diffusion model remains frozen with no additional language conditioning; the meta-learning gradient is enough to produce both stylistic variations and concept compositions that the base model does not generate on its own.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。