Content-Based Search for Deep Generative Models

TL;DR

This paper introduces a content-based search method for deep generative models, enabling users to find models matching a query across different modalities by formulating an optimization problem and employing contrastive learning, outperforming baselines on a new benchmark.

Contribution

The paper proposes a novel content-based model search framework for generative models, including a probabilistic formulation and a contrastive learning approach for multi-modal queries.

Findings

Outperforms baseline methods on the Generative Model Zoo benchmark.

Effective retrieval across image, sketch, and text modalities.

Introduces a new benchmark dataset for model retrieval tasks.

Abstract

The growing proliferation of customized and pretrained generative models has made it infeasible for a user to be fully cognizant of every model in existence. To address this need, we introduce the task of content-based model search: given a query and a large set of generative models, finding the models that best match the query. As each generative model produces a distribution of images, we formulate the search task as an optimization problem to select the model with the highest probability of generating similar content as the query. We introduce a formulation to approximate this probability given the query from different modalities, e.g., image, sketch, and text. Furthermore, we propose a contrastive learning framework for model retrieval, which learns to adapt features for various query modalities. We demonstrate that our method outperforms several baselines on Generative Model Zoo, a new benchmark we create for the model retrieval task.