A Brief Review of the Performance of Energy-Based Models in Embedded Devices

Today, embedded devices are widely used in both personal and industrial applications such as monitoring and control, visual data processing and many more. ARM-based processors are often preferred in embedded devices and especially in single-board platforms. The increased processing power and capabilities of these processors are highly effective in increasing the efficiency of these processes. Energy-based models are one of the many machine learning methods that can be used in these devices and are becoming increasingly popular. In this paper, the performance of the Restricted Boltzmann machine, an energy-based modeling approach, on different ARM-based processors is analyzed and its potential applications are discussed.