Overview of Convolutional Neural Networks
Arslan Munir, Joonho Kong, Mahmood Azhar Qureshi · 2023
Convolutional neural networks (CNNs) are a specialized type of deep neural networks (DNNs) that utilize a mathematical operation called convolution instead of general matrix multiplication in at least one of their layers. This chapter gives an overview of the architecture of CNNs and explores the different layers that these neural networks comprised of focusing on convolutional layers, pooling layers, and fully connected layers. The chapter further discusses different parameters and hyperparameters of CNNs. The chapter then highlights some of the most popular high-accuracy CNN models and the popular datasets upon which these CNNs operate. Finally, the chapter reviews some of the architectural and algorithmic techniques for efficient processing of high-accuracy CNN models on edge devices.