Generative Adversarial Network Training is a Continual Learning Problem

Generative Adversarial Networks (GANs) have proven to be a powerful framework for learning to draw samples from complex distributions. However, GANs are also notoriously difficult to train, with mode collapse and oscillations a common problem. We hypothesize that this is at least in part due to the evolution of the generator distribution and the catastrophic forgetting tendency of neural networks, which leads to the discriminator losing the ability to remember synthesized samples from previous instantiations of the generator. Recognizing this, our contributions are twofold. First, we show that GAN training makes for a more interesting and realistic benchmark for continual learning methods evaluation than some of the more canonical datasets. Second, we propose leveraging continual learning techniques to augment the discriminator, preserving its ability to recognize previous generator samples. We show that the resulting methods add only a light amount of computation, involve minimal changes to the model, and result in better overall performance on the examined image and text generation tasks.

[1]  Kevin Lin,et al.  Adversarial Ranking for Language Generation , 2017, NIPS.

[2]  Ben Poole,et al.  Categorical Reparameterization with Gumbel-Softmax , 2016, ICLR.

[3]  Charles A. Sutton,et al.  VEEGAN: Reducing Mode Collapse in GANs using Implicit Variational Learning , 2017, NIPS.

[4]  Yoshua Bengio,et al.  Generative Adversarial Networks , 2014, ArXiv.

[5]  Lantao Yu,et al.  SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient , 2016, AAAI.

[6]  Yoshua Bengio,et al.  An Empirical Investigation of Catastrophic Forgeting in Gradient-Based Neural Networks , 2013, ICLR.

[7]  Zhi Chen,et al.  Adversarial Feature Matching for Text Generation , 2017, ICML.

[8]  Xinlei Chen,et al.  Microsoft COCO Captions: Data Collection and Evaluation Server , 2015, ArXiv.

[9]  Jaakko Lehtinen,et al.  Progressive Growing of GANs for Improved Quality, Stability, and Variation , 2017, ICLR.

[10]  Shane Legg,et al.  Human-level control through deep reinforcement learning , 2015, Nature.

[11]  Simon Osindero,et al.  Conditional Generative Adversarial Nets , 2014, ArXiv.

[12]  Geoffrey E. Hinton,et al.  Distilling the Knowledge in a Neural Network , 2015, ArXiv.

[13]  Truyen Tran,et al.  On catastrophic forgetting and mode collapse in Generative Adversarial Networks , 2018, ArXiv.

[14]  Razvan Pascanu,et al.  Overcoming catastrophic forgetting in neural networks , 2016, Proceedings of the National Academy of Sciences.

[15]  Behnam Neyshabur,et al.  Stabilizing GAN Training with Multiple Random Projections , 2017, ArXiv.

[16]  Philip H. S. Torr,et al.  Multi-agent Diverse Generative Adversarial Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[17]  Tomas Pfister,et al.  Learning from Simulated and Unsupervised Images through Adversarial Training , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[18]  Sridhar Mahadevan,et al.  Generative Multi-Adversarial Networks , 2016, ICLR.

[19]  Sebastian Nowozin,et al.  The Numerics of GANs , 2017, NIPS.

[20]  Michael McCloskey,et al.  Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem , 1989 .

[21]  Yong Yu,et al.  Long Text Generation via Adversarial Training with Leaked Information , 2017, AAAI.

[22]  Aaron C. Courville,et al.  Improved Training of Wasserstein GANs , 2017, NIPS.

[23]  Trung Le,et al.  MGAN: Training Generative Adversarial Nets with Multiple Generators , 2018, ICLR.

[24]  Yarin Gal,et al.  Towards Robust Evaluations of Continual Learning , 2018, ArXiv.

[25]  Wojciech Zaremba,et al.  Improved Techniques for Training GANs , 2016, NIPS.

[26]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[27]  J. Zico Kolter,et al.  Gradient descent GAN optimization is locally stable , 2017, NIPS.

[28]  Yuichi Yoshida,et al.  Spectral Normalization for Generative Adversarial Networks , 2018, ICLR.

[29]  R Ratcliff,et al.  Connectionist models of recognition memory: constraints imposed by learning and forgetting functions. , 1990, Psychological review.

[30]  Han Liu,et al.  Continual Learning in Generative Adversarial Nets , 2017, ArXiv.

[31]  Fernando A. Mujica,et al.  An Empirical Evaluation of Deep Learning on Highway Driving , 2015, ArXiv.

[32]  Salim Roukos,et al.  Bleu: a Method for Automatic Evaluation of Machine Translation , 2002, ACL.

[33]  Lawrence Carin,et al.  Automatic threat recognition of prohibited items at aviation checkpoint with x-ray imaging: a deep learning approach , 2018, Defense + Security.

[34]  Andrew Gordon Wilson,et al.  Bayesian GAN , 2017, NIPS.

[35]  Yoshua Bengio,et al.  Mode Regularized Generative Adversarial Networks , 2016, ICLR.

[36]  Sepp Hochreiter,et al.  GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium , 2017, NIPS.

[37]  Ole Winther,et al.  Autoencoding beyond pixels using a learned similarity metric , 2015, ICML.

[38]  Alan Ritter,et al.  Adversarial Learning for Neural Dialogue Generation , 2017, EMNLP.

[39]  Surya Ganguli,et al.  Continual Learning Through Synaptic Intelligence , 2017, ICML.