Seeking the Unfamiliar but Memorable: Conceptual Creativity as Meta-Learning

TL;DR

This paper introduces a meta-learning framework where a Creator generates novel stimuli, and an Appraiser quickly learns from them to produce unfamiliar yet learnable concepts, enhancing creative outputs.

Contribution

It formalizes a Creator-Appraiser meta-learning paradigm and demonstrates its effectiveness with diffusion models and autoencoder/CLIP Appraisers for creative concept generation.

Findings

Diffusion models can produce stylistic variations without additional conditioning.

Meta-learning enables creation of unfamiliar, learnable concepts.

Framework improves diversity and novelty in generated stimuli.

Abstract

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.