Deep learning on mobile devices: a review

Abstract

Recent breakthroughs in deep learning and artificial intelligence technologies have enabled numerous mobile applications. While traditional computation paradigms rely on mobile sensing and cloud computing, deep learning implemented on mobile devices provides several advantages. These advantages include low communication bandwidth, small cloud computing resource cost, quick response time, and improved data privacy. Research and development of deep learning on mobile and embedded devices has recently attracted much attention. This paper provides a timely review of this fast-paced field to give the researcher, engineer, practitioner, and graduate student a quick grasp on the recent advancements of deep learning on mobile devices. In this paper, we discuss hardware architectures for mobile deep learning, including Field Programmable Gate Arrays (FPGA), Application Specific Integrated Circuit (ASIC), and recent mobile Graphic Processing Units (GPUs). We present Size, Weight, Area and Power (SWAP) considerations and their relation to algorithm optimizations, such as quantization, pruning, compression, and approximations that simplify computation while retaining performance accuracy. We cover existing systems and give a state-of-the-industry review of TensorFlow, MXNet, Mobile AI Compute Engine (MACE), and Paddle-mobile deep learning platform. We discuss resources for mobile deep learning practitioners, including tools, libraries, models, and performance benchmarks. We present applications of various mobile sensing modalities to industries, ranging from robotics, healthcare and multimedia, biometrics to autonomous drive and defense. We address the key deep learning challenges to overcome, including low quality data, and small training/adaptation data sets. In addition, the review provides numerous citations and links to existing code bases implementing various technologies. These resources lower the user's barrier to entry into the field of mobile deep learning.

PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning

“…In the example (shown on the left part of Figure 11), the kernel value 1 requires the elements in the first two rows of the input matrix while value 2 requires the second and third rows. The elements in the second row [7,8,9,10] are loaded twice (from cache to register). PatDNN eliminates this load redundancy by explicitly reusing the (SIMD) registers that already hold the required data (like the second row in the above example).…”

“…After obtaining DNN models trained with a huge amount of data, they can be deployed for inference, perception and control tasks in various autonomous systems and internet-of-things (IoT) applications. Recently, along arXiv:2001.00138v4 [cs.LG] 22 Jan 2020 with the rapid emergence of high-end mobile devices 1 , executing DNNs on mobile platforms gains popularity and is quickly becoming the mainstream [9,28,30,43,63] for broad applications such as sensor nodes, wireless access points, smartphones, wearable devices, video streaming, augmented reality, robotics, unmanned vehicles, smart health devices, etc. [2,3,29,46,50].…”

Intelligent Mobile Applications: A Systematic Mapping Study

“…Like ER, VR are also pertinent and can be considered in future SLR studies. Other research types must be ignored in subsequent work [14]. e codominance of SP, VR, and PP to the detriment of ER reflects that the majority of proposed solutions are not implemented or experimented in real context.…”

“…Recently, the introduction of machine learning techniques in the creation of intelligent mobile apps is the subject of several research works of which the number keeps increasing year by year. is new orientation towards mobile applications is encouraged by the growth in performance of smartphones in terms of CPU power, RAM capacity, and energy storage, and also it is due to advances in the field of cloud computing that provide an on-demand cloud services of data storage and computing power [14]. us, to structure this research axis and to facilitate the information extraction about various available publications, it becomes imperative to carry out a literature study.…”

Level Scaling and Pulse Regulating Methods to Mitigate Cycle-to-cycle Variation in Memristor Based Learning System

“…Recent years have witnessed significant progress in edge devices and wireless sensor networks, creating unprecedented opportunities to deploy deep learning and artificial intelligence (AI) technologies in IoT, while significantly adding calculation burdens in edges 2 . However, edges consisting of mobile devices and embedded systems usually have limited resources and power, especially when they are used for real-time applications, such resource and power deficiency will results in recognition and prediction accuracy loss in a learning system and even malfunctions in IoT [1][2][3][4][5][6] .…”