Home // BUSTECH 2023, The Thirteenth International Conference on Business Intelligence and Technology // View article

Supervised Classification with Deep Graph CNN

Authors:
Mathias Tiberghien
Rakia Jaziri

Keywords: convolution neural network; image processing; classification; neuronal network architecture; deep learning; incremental learning

Abstract:
Convolution neural networks (CNNs) have performed remarkably well in recent decades and become essential for classification tasks based on images or voice. However, this paper addresses some of their limitations in terms of generalization and examines some well-known CNNs architectures that try to create hierarchical structures based on graph databases. The contribution of this work is to present a structure where the convolution blocks are distributed in nodes, the relations between each node being articulated using a data partitioner. This exponentially multiplies the number of models depending on the depth of the graph and the number of partitions, but it keeps track of the hierarchical relationships between each node.

Pages: 1 to 6

Copyright: Copyright (c) IARIA, 2023

Publication date: June 26, 2023

Published in: conference

ISSN: 2308-4391

ISBN: 978-1-68558-051-3

Location: Nice, France

Dates: from June 26, 2023 to June 30, 2023