CAM-GAN: Continual Adaptation Modules for Generative Adversarial Networks

Varshney, Sakshi and Verma, Vinay Kumar and Srijith, P K (2021) CAM-GAN: Continual Adaptation Modules for Generative Adversarial Networks. In: 35th Conference on Neural Information Processing Systems, NeurIPS 2021, 6 December 2021 through 14 December 2021, Virtual, Online.

Full text not available from this repository. (Request a copy)

Abstract

We present a continual learning approach for generative adversarial networks (GANs), by designing and leveraging parameter-efficient feature map transformations. Our approach is based on learning a set of global and task-specific parameters. The global parameters are fixed across tasks whereas the task-specific parameters act as local adapters for each task, and help in efficiently obtaining task-specific feature maps. Moreover, we propose an element-wise addition of residual bias in the transformed feature space, which further helps stabilize GAN training in such settings. Our approach also leverages task similarities based on the Fisher information matrix. Leveraging this knowledge from previous tasks significantly improves the model performance. In addition, the similarity measure also helps reduce the parameter growth in continual adaptation and helps to learn a compact model. In contrast to the recent approaches for continually-learned GANs, the proposed approach provides a memory-efficient way to perform effective continual data generation. Through extensive experiments on challenging and diverse datasets, we show that the feature-map-transformation approach outperforms state-of-the-art methods for continually-learned GANs, with substantially fewer parameters. The proposed method generates high-quality samples that can also improve the generative-replay-based continual learning for discriminative tasks. © 2021 Neural information processing systems foundation. All rights reserved.

[error in script]
IITH Creators:
IITH CreatorsORCiD
Srijith, P Khttps://orcid.org/0000-0002-2820-0835
Item Type: Conference or Workshop Item (Paper)
Additional Information: Sakshi Varshney acknowledges the support from DST ICPS and Visvesvaraya fellowship. The portion of this research performed at Duke University was supported under the DARPA L2M program. PR acknowledges support from Visvesvaraya Young Faculty Fellowship.
Uncontrolled Keywords: Adaptation module; Continual learning; Feature map; Feature space; Fisher information matrices; Global parameters; Learning approach; Modeling performance; Network training; Similarity measure
Subjects: Computer science
Divisions: Department of Computer Science & Engineering
Depositing User: . LibTrainee 2021
Date Deposited: 08 Aug 2022 06:58
Last Modified: 08 Aug 2022 06:58
URI: http://raiith.iith.ac.in/id/eprint/10140
Publisher URL:
Related URLs:

Actions (login required)

View Item View Item
Statistics for RAIITH ePrint 10140 Statistics for this ePrint Item