Which generative model is the most suitable for Continual Learning? This
paper aims at evaluating and comparing generative models on disjoint sequential
image generation tasks. We investigate how several models learn and forget,
considering various strategies: rehearsal, regularization, generative replay
and fine-tuning. We used two quantitative metrics to estimate the generation
quality and memory ability. We experiment with sequential tasks on three
commonly used benchmarks for Continual Learning (MNIST, Fashion MNIST and
CIFAR10). We found that among all models, the original GAN performs best and
among Continual Learning strategies, generative replay outperforms all other
methods. Even if we found satisfactory combinations on MNIST and Fashion MNIST,
training generative models sequentially on CIFAR10 is particularly instable,
and remains a challenge. Our code is available online
\footnote{https://github.com/TLESORT/Generative\_Continual\_Learning}.

本文比较评估了不同生成模型在连续学习时的适用性，主要研究了不同策略下生成模型的记忆及生成质量表现，结论为原始 GAN 模型表现最优，而 generative replay 策略最优。