SeNA-CNN: Overcoming Catastrophic Forgetting in Convolutional Neural Networks by Selective Network Augmentation

TL;DR

SeNA-CNN introduces a selective network augmentation method that enables convolutional neural networks to learn new tasks without forgetting previous ones, outperforming some existing methods in certain scenarios.

Contribution

The paper proposes a novel selective network augmentation approach to mitigate catastrophic forgetting in CNNs during lifelong learning.

Findings

SeNA-CNN outperforms Learning without Forgetting in some scenarios.

SeNA-CNN can be preferable to isolated training in certain situations.

The method preserves old task performance while learning new tasks without data access.

Abstract

Lifelong learning aims to develop machine learning systems that can learn new tasks while preserving the performance on previous learned tasks. In this paper we present a method to overcome catastrophic forgetting on convolutional neural networks, that learns new tasks and preserves the performance on old tasks without accessing the data of the original model, by selective network augmentation. The experiment results showed that SeNA-CNN, in some scenarios, outperforms the state-of-art Learning without Forgetting algorithm. Results also showed that in some situations it is better to use SeNA-CNN instead of training a neural network using isolated learning.