Kernel Mean Matching for Content Addressability of GANs

TL;DR

This paper introduces a kernel mean matching method to enable content-addressability in GANs, allowing users to generate images similar to a specified set without retraining the model.

Contribution

It presents a novel, retraining-free approach to add content-addressability to any implicit generative model using kernel mean matching.

Findings

Successfully generates images consistent with input sets

Maintains high image quality comparable to original GANs

Applicable to various high-dimensional image datasets

Abstract

We propose a novel procedure which adds "content-addressability" to any given unconditional implicit model e.g., a generative adversarial network (GAN). The procedure allows users to control the generative process by specifying a set (arbitrary size) of desired examples based on which similar samples are generated from the model. The proposed approach, based on kernel mean matching, is applicable to any generative models which transform latent vectors to samples, and does not require retraining of the model. Experiments on various high-dimensional image generation problems (CelebA-HQ, LSUN bedroom, bridge, tower) show that our approach is able to generate images which are consistent with the input set, while retaining the image quality of the original model. To our knowledge, this is the first work that attempts to construct, at test time, a content-addressable generative model from a trained marginal model.