An Early Resource Characterization of Deep Learning on Wearables, Smartphones and Internet-of-Things Devices

By Nicholas Donald Lane, Sourav Bhattacharya, Petko Georgiev, Claudio Forlivesi, F. Kawsar. 2015

220 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Learning for IoT Big Data and Streaming Analytics: A Survey (2017, background): “In some cases, the available resources are even not sufficient for running a pre-trained DL algorithm for inference tasks [80].”
Accelerating Mobile Audio Sensing Algorithms through On-Chip GPU Offloading (2017, background): “These applications are capable of deep inferences about user behavior, often require continuous sensor monitoring, and boast highly sophisticated inference algorithms that easily strain scarce mobile resources(battery, memory and computation) [41].”
DeepX: A Software Accelerator for Low-Power Deep Learning Inference on Mobile Devices (2016, background): “For example 93% of the total memory needed for AlexNet is concentrated only in two fully connected layers [44].”
Context-Aware Computing, Learning, and Big Data in Internet of Things: A Survey (2018, background): “Lane et al. [145] researched the use of deep learning implementations to process data from IoT devices, such as wearables and smart phones.”
Deep Learning in Mobile and Wireless Networking: A Survey (2018, background): “Advanced deep learning practices provide a powerful solution for app-level data mining, as they demonstrate better precision and higher robustness in IoT applications [380].”
A Survey of Wearable Devices and Challenges (2017, background): “As a result, number of work evaluated the possibility of using deep learning models in wearables [274], [275] and proposed modified versions of deep learning models that are suitable for wearables [276].”
MaxNVM: Maximizing DNN Storage Density and Inference Efficiency with Sparse Encoding and Error Mitigation (2019, background): “DNNs are in use everywhere from self-driving cars to wireless sensor nodes and implanted medical devices [16, 40, 43, 50, 52].”
Survey on Deep Neural Networks in Speech and Vision Systems (2019, methodology): “[120] have also performed an initial study using two popular deep learning models: CNN and fully connected deep feed-forward networks, to analyze audio and image data on three hardware platforms: Qualcomm Snapdragon 800, Intel Edison, and Nvidia Tegra K1 as these are commonly used in wearable and mo”
Toward Massive Machine Type Communications in Ultra-Dense Cellular IoT Networks: Current Issues and Machine Learning-Assisted Solutions (2018, methodology): “In this regard, authors in [37] investigated the applicability of DL models on different hardware devices including Intel Edison used in wearables, Snapdragon 800 employed in some models of smartphones and Nvidia Tegra K1 used in IoTenabled vehicles by taking measurements of energy consumption, time”
Estimation of energy consumption in machine learning (2019, methodology): “Some awareness in energy consumption is starting to arise, originating from a fewmachine learning research groups [12,14,47,61] and challenges such as The Low Power Image Recognition Challenge (LPIRC) [26].”
Learning-Based Computation Offloading for IoT Devices With Energy Harvesting (2017, methodology): “This scheme can be implemented on IoT platforms such as Qualcomm Snapdragon 800 and Nvidia Tegra K1 that support for deep learning [19], [20].”
Cellular, Wide-Area, and Non-Terrestrial IoT: A Survey on 5G Advances and the Road Toward 6G (2021, background): “Even worse, resources may not be sufficient for the test phase where a pre-trained model is run on the device [349].”
Efficient Acceleration of Deep Learning Inference on Resource-Constrained Edge Devices: A Review (2023, background): “Similarly, researchers also investigated DL training in wearable [347] and embedded [348] devices.”
Unbox the black-box for the medical explainable AI via multi-modal and multi-centre data fusion: A mini-review, two showcases and beyond (2021, unknown): “In general, deep learning based AI algorithms have been developed for image-based classification [43], diagnosis [44, 45, 46] and prognosis [47, 48], genome interpretation [49], biomarker discovery [50, 51], monitoring by wearable life-logging devices [52], and automated robotic surgery [53] to enha”
DeepThings: Distributed Adaptive Deep Learning Inference on Resource-Constrained IoT Edge Clusters (2018, background): “embedded devices requires large computational resources and memory footprints [3] that are usually not available in IoT computing platforms.”