Memory Replay GANs: learning to generate images from new categories without forgetting
This addresses the issue of forgetting in generative models for sequential learning, which is incremental as it builds on existing discriminative methods.
The paper tackles the problem of catastrophic forgetting in generative adversarial networks (GANs) when learning new categories sequentially, proposing Memory Replay GANs (MeRGANs) to mitigate forgetting and generate competitive images from previous categories.
Previous works on sequential learning address the problem of forgetting in discriminative models. In this paper we consider the case of generative models. In particular, we investigate generative adversarial networks (GANs) in the task of learning new categories in a sequential fashion. We first show that sequential fine tuning renders the network unable to properly generate images from previous categories (i.e. forgetting). Addressing this problem, we propose Memory Replay GANs (MeRGANs), a conditional GAN framework that integrates a memory replay generator. We study two methods to prevent forgetting by leveraging these replays, namely joint training with replay and replay alignment. Qualitative and quantitative experimental results in MNIST, SVHN and LSUN datasets show that our memory replay approach can generate competitive images while significantly mitigating the forgetting of previous categories.