Diffusion Autoencoders for Few-shot Image Generation in Hyperbolic Space

2024-11-27

arXiv.org

AI Summary

Few-shot image generation faces challenges in simultaneously ensuring class consistency and image diversity, while maintaining fine-grained, interpretable attribute controllability. To address this, we propose the first diffusion-based autoencoder framework embedded in hyperbolic space, which leverages hyperbolic geometry to model the hierarchical semantic structure of image–text pairs. By modulating the radius of the Poincaré disk, our method enables fine-grained, interpretable control over semantic diversity. The architecture integrates semantic priors from pretrained multimodal models (e.g., CLIP), a variational encoder–decoder backbone, and a hyperbolic diffusion process—enabling high-fidelity, diverse, and text-guided generation from only a few examples. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods across multiple few-shot benchmarks, achieving, for the first time, a unified balance among generation quality, controllability, and interpretability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Few-shot image generation aims to generate diverse and high-quality images for an unseen class given only a few examples in that class. However, existing methods often suffer from a trade-off between image quality and diversity while offering limited control over the attributes of newly generated images. In this work, we propose Hyperbolic Diffusion Autoencoders (HypDAE), a novel approach that operates in hyperbolic space to capture hierarchical relationships among images and texts from seen categories. By leveraging pre-trained foundation models, HypDAE generates diverse new images for unseen categories with exceptional quality by varying semantic codes or guided by textual instructions. Most importantly, the hyperbolic representation introduces an additional degree of control over semantic diversity through the adjustment of radii within the hyperbolic disk. Extensive experiments and visualizations demonstrate that HypDAE significantly outperforms prior methods by achieving a superior balance between quality and diversity with limited data and offers a highly controllable and interpretable generation process.

Problem

• Balancing category consistency and image diversity
• Providing control over attributes of generated images
• Generating hierarchical few-shot images with limited data

Innovation

• Hyperbolic space captures hierarchical image relationships
• Adjusts radii in hyperbolic disk for semantic diversity control
• Leverages pre-trained models for few-shot generation quality