Lifelong GAN: Continual Learning for Conditional Image Generation

Mengyao Zhai, Lei Chen, Fred Tung, Jiawei He, Megha Nawhal, Greg Mori

2019 · 22 citations · 26 references

DOIFull text

Open access

Concepts

TL;DR

Lifelong learning is challenging for deep neural networks because catastrophic forgetting causes a network to lose previously learned task performance when trained on new tasks. The study aims to extend generative models to new conditional generation tasks without forgetting prior tasks, proposing a generic continual learning framework beyond label‑conditioned image generation. Lifelong GAN uses knowledge distillation to transfer knowledge from previous networks to the new network, enabling continual learning across different conditional image generation settings. Experiments on image‑conditioned and label‑conditioned tasks demonstrate Lifelong GAN’s generality and effectiveness, achieving successful lifelong conditional image generation.

Abstract

Lifelong learning is challenging for deep neural networks due to their susceptibility to catastrophic forgetting. Catastrophic forgetting occurs when a trained network is not able to maintain its ability to accomplish previously learned tasks when it is trained to perform new tasks. We study the problem of lifelong learning for generative models, extending a trained network to new conditional generation tasks without forgetting previous tasks, while assuming access to the training data for the current task only. In contrast to state-of-the-art memory replay based approaches which are limited to label-conditioned image generation tasks, a more generic framework for continual learning of generative models under different conditional image generation settings is proposed in this paper. Lifelong GAN employs knowledge distillation to transfer learned knowledge from previous networks to the new network. This makes it possible to perform image-conditioned generation tasks in a lifelong learning setting. We validate Lifelong GAN for both image-conditioned and label-conditioned generation tasks, and provide qualitative and quantitative results to show the generality and effectiveness of our method.

References

26