Flow Plugin Network for conditional generation

TL;DR

This paper introduces a Flow Plugin Network (FPN) that allows conditional generation of objects with specific attributes without retraining the base generative model, leveraging normalizing flow models.

Contribution

The novel FPN approach enables attribute-controlled generation without retraining, improving flexibility and efficiency in generative modeling.

Findings

Enables attribute-specific sample generation without retraining

Uses normalizing flow models as a plugin for conditional control

Reduces computational resources needed for controlled generation

Abstract

Generative models have gained many researchers' attention in the last years resulting in models such as StyleGAN for human face generation or PointFlow for the 3D point cloud generation. However, by default, we cannot control its sampling process, i.e., we cannot generate a sample with a specific set of attributes. The current approach is model retraining with additional inputs and different architecture, which requires time and computational resources. We propose a novel approach that enables to a generation of objects with a given set of attributes without retraining the base model. For this purpose, we utilize the normalizing flow models - Conditional Masked Autoregressive Flow and Conditional Real NVP, as a Flow Plugin Network (FPN).