THE PERFORMANCE ANALYSIS OF DEEP LEARNING NEURAL NETWORKS ON MOBILE DATA

Abstract

In recent years, the genuine feasibility of training deep learning models on mobile data has been
facilitated by the advancements in processing power of mobile technology and the advantages it offers in terms
of enabling enhanced user experiences. The predominant focus of contemporary research in mobile machine
learning pertains to the inference phase within the realm of deep learning model creation. The exploration of
performance characterization for training deep learning models on mobile devices is mostly unexplored, despite
its vital importance in the development and deployment of such models. Deep learning approaches have shown
superior performance compared to earlier state-of-the-art machine learning methods, particularly in the domain
of computer vision. In addition to using deep learning methodologies for tasks like as classification and
segmentation, it is also used for the purpose of training neural networks via the utilization of photographic data.
As the accuracy of contemporary deep learning models continues to advance, there has been a corresponding
increase in their size and depth, which enables them to effectively address certain tasks