|   | 
Details
   web
Record Links
Author (up) Chenshen Wu; Luis Herranz; Xialei Liu; Joost Van de Weijer; Bogdan Raducanu edit   pdf
openurl 
Title Memory Replay GANs: Learning to Generate New Categories without Forgetting Type Conference Article
Year 2018 Publication 32nd Annual Conference on Neural Information Processing Systems Abbreviated Journal  
Volume Issue Pages 5966-5976  
Keywords  
Abstract 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 (ie 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.  
Address Montreal; Canada; December 2018  
Corporate Author Thesis  
Publisher Place of Publication Editor  
Language Summary Language Original Title  
Series Editor Series Title Abbreviated Series Title  
Series Volume Series Issue Edition  
ISSN ISBN Medium  
Area Expedition Conference NIPS  
Notes LAMP; 600.106; 600.109; 602.200; 600.120;MV;OR;CIC Approved no  
Call Number Admin @ si @ WHL2018 Serial 3249  
Permanent link to this record