CLoG: Benchmarking Continual Learning of Image Generation Models

TL;DR

This paper introduces a new benchmarking framework for Continual Learning of Generative models (CLoG), adapting existing CL methods to generative tasks and providing comprehensive benchmarks to advance research in this area.

Contribution

It systematically identifies challenges of CLoG, adapts three CL methodologies for generative models, and releases a benchmark suite and codebase for future research.

Findings

Adapted replay, regularization, and parameter-isolation methods for CLoG

Created diverse benchmarks covering broad tasks in CLoG

Provided insights to guide future CLoG method development

Abstract

Continual Learning (CL) poses a significant challenge in Artificial Intelligence, aiming to mirror the human ability to incrementally acquire knowledge and skills. While extensive research has focused on CL within the context of classification tasks, the advent of increasingly powerful generative models necessitates the exploration of Continual Learning of Generative models (CLoG). This paper advocates for shifting the research focus from classification-based CL to CLoG. We systematically identify the unique challenges presented by CLoG compared to traditional classification-based CL. We adapt three types of existing CL methodologies, replay-based, regularization-based, and parameter-isolation-based methods to generative tasks and introduce comprehensive benchmarks for CLoG that feature great diversity and broad task coverage. Our benchmarks and results yield intriguing insights that can be valuable for developing future CLoG methods. Additionally, we will release a codebase designed to facilitate easy benchmarking and experimentation in CLoG publicly at https://github.com/linhaowei1/CLoG. We believe that shifting the research focus to CLoG will benefit the continual learning community and illuminate the path for next-generation AI-generated content (AIGC) in a lifelong learning paradigm.