Mobile Neural Networks and CNNs

The computation costs required by deep learning networks have always been a concern for expansion. Millions of multiplication operations are required to run an inference. This has limited the practical use of developed convolutional neural network (CNN) models. The mobile neural network provides a breakthrough to this problem. They are super small and computationally light deep learning networks, and achieve performance that's equivalent to their original counterparts. Mobile neural networks are just CNNs that have been modified to have far fewer parameters, which means they are consuming less memory. This way, they are capable of working on mobile devices with limited memory and processing power. Hence, mobile neural networks are playing a crucial role in making CNNs work for real-time applications. In this chapter, we...