Context-Aware Compilation of DNN Training Pipelines across Edge and Cloud

2021 ◽  
Vol 5 (4) ◽  
pp. 1-27
Author(s):  
Dixi Yao ◽  
Liyao Xiang ◽  
Zifan Wang ◽  
Jiayu Xu ◽  
Chao Li ◽  
...  
Keyword(s):  
Cloud Model ◽  
Limited Resource ◽  
Parallel Execution ◽  
Integrated System ◽  
Context Aware ◽  
Training Models ◽  
Classification Image ◽  
Model Training ◽  
Iot Devices ◽  
Additional Memory

Empowered by machine learning, edge devices including smartphones, wearable, and IoT devices have become growingly intelligent, raising conflicts with the limited resource. On-device model personalization is particularly hard as training models on edge devices is highly resource-intensive. In this work, we propose a novel training pipeline across the edge and the cloud, by taking advantage of the powerful cloud while keeping data local at the edge. Highlights of the design incorporate the parallel execution enabled by our feature replay, reduced communication cost by our error-feedback feature compression, as well as the context-aware deployment decision engine. Working as an integrated system, the proposed pipeline training framework not only significantly speeds up training, but also incurs little accuracy loss or additional memory/energy overhead. We test our system in a variety of settings including WiFi, 5G, household IoT, and on different training tasks such as image/text classification, image generation, to demonstrate its advantage over the state-of-the-art. Experimental results show that our system not only adapts well to, but also draws on the varying contexts, delivering a practical and efficient solution to edge-cloud model training.

scholarly journals Using Machine Learning Methods Incorporating Individual Reader Annotations to Classify Paediatric Chest Radiographs in Epidemiological Studies

2021 ◽  
Vol 6 ◽  
pp. 309
Author(s):  
Paul Mwaniki ◽  
Timothy Kamanu ◽  
Samuel Akech ◽  
M. J. C Eijkemans
Keyword(s):  
Machine Learning ◽  
Epidemiological Studies ◽  
Chest Radiographs ◽  
World Health ◽  
Data Sets ◽  
Learning Models ◽  
Middle Income ◽  
Training Models ◽  
Model Training ◽  
Machine Learning Models

Introduction: Epidemiological studies that involve interpretation of chest radiographs (CXRs) suffer from inter-reader and intra-reader variability. Inter-reader and intra-reader variability hinder comparison of results from different studies or centres, which negatively affects efforts to track the burden of chest diseases or evaluate the efficacy of interventions such as vaccines. This study explores machine learning models that could standardize interpretation of CXR across studies and the utility of incorporating individual reader annotations when training models using CXR data sets annotated by multiple readers. Methods: Convolutional neural networks were used to classify CXRs from seven low to middle-income countries into five categories according to the World Health Organization's standardized methodology for interpreting paediatric CXRs. We compared models trained to predict the final/aggregate classification with models trained to predict how each reader would classify an image and then aggregate predictions for all readers using unweighted mean. Results: Incorporating individual reader's annotations during model training improved classification accuracy by 3.4% (multi-class accuracy 61% vs 59%). Model accuracy was higher for children above 12 months of age (68% vs 58%). The accuracy of the models in different countries ranged between 45% and 71%. Conclusions: Machine learning models can annotate CXRs in epidemiological studies reducing inter-reader and intra-reader variability. In addition, incorporating individual reader annotations can improve the performance of machine learning models trained using CXRs annotated by multiple readers.

scholarly journals A context-aware encryption protocol suite for edge computing-based IoT devices

2019 ◽  
Vol 76 (4) ◽  
pp. 2548-2567
Author(s):  
Zaineb Dar ◽  
Adnan Ahmad ◽  
Farrukh Aslam Khan ◽  
Furkh Zeshan ◽  
Razi Iqbal ◽  
...  
Keyword(s):  
Edge Computing ◽  
Context Aware ◽  
Iot Devices

scholarly journals Methodology for Collecting a Training Dataset for an Intrusion Detection Model

2021 ◽  
Vol 33 (5) ◽  
pp. 83-104
Author(s):  
Aleksandr Igorevich Getman ◽  
Maxim Nikolaevich Goryunov ◽  
Andrey Georgievich Matskevich ◽  
Dmitry Aleksandrovich Rybolovlev
Keyword(s):  
Attack Detection ◽  
Training Data ◽  
Training Dataset ◽  
Training Models ◽  
The Public ◽  
Detection Model ◽  
Computer Attacks ◽  
Model Training ◽  
Public Datasets

The paper discusses the issues of training models for detecting computer attacks based on the use of machine learning methods. The results of the analysis of publicly available training datasets and tools for analyzing network traffic and identifying features of network sessions are presented sequentially. The drawbacks of existing tools and possible errors in the datasets formed with their help are noted. It is concluded that it is necessary to collect own training data in the absence of guarantees of the public datasets reliability and the limited use of pre-trained models in networks with characteristics that differ from the characteristics of the network in which the training traffic was collected. A practical approach to generating training data for computer attack detection models is proposed. The proposed solutions have been tested to evaluate the quality of model training on the collected data and the quality of attack detection in conditions of real network infrastructure.

scholarly journals Context Aware Data Fusion on Massive IOT Data in Dynamic IOT Analytics

Webology ◽  
2020 ◽  
Vol 17 (2) ◽  
pp. 957-970
Author(s):  
S.S. Saranya ◽  
Dr.N. Sabiyath Fatima
Keyword(s):  
Data Fusion ◽  
Data Cleaning ◽  
Research Work ◽  
Context Aware ◽  
Multiple Sources ◽  
Combine Data ◽  
Optimal Outcomes ◽  
Iot Devices ◽  
Noise Reduction Algorithm ◽  
Combined Data

Educational Data management is a critical task for the researchers due to mammoth data generated by sensors and IoT (Internet of Things) devices. Managing this huge volume of data, cleaning this data from impurities is an inherent need. DF (Data Fusion) processes combine data from multiple sources based on their similarity for an easy management. DF processes focus on many factors like nature of data and application that uses that data. Many DFAs (Data Fusion approaches) have been proposed without detailing on the context for integrating data in fusion tasks. This work attempts to cover this gap of context’s relevance by proposing a technique CDFT (Context aware Data Fusion technique). In this research work, initially data from IoT devices will be gathered and pre-processed to make it clear for the fusion processing. In this work, boundary based noise reduction algorithm is introduced for data pre-processing which attempts to label the unlabelled attributes in the data’s that are gathered, so that data fusion can be done accurately. After pre-processing Context aware data fusion is performed which will combine the data’s from multiple IoT devices together with the concern of context. Finally this combined data will be learnt using the convolution neural network for data fusion performance checking. The proposed CDFT is simulated on Matlab whose results prove that the proposed technique obtains optimal outcomes.

scholarly journals On the Sensitivity of TanDEM-X-Observations to Boreal Forest Structure

2019 ◽  
Vol 11 (14) ◽  
pp. 1644 ◽  
Author(s):  
Jan I. H. Askne ◽  
Henrik J. Persson ◽  
Lars M. H. Ulander
Keyword(s):  
Boreal Forest ◽  
Cloud Model ◽  
Tree Height ◽  
Airborne Lidar ◽  
Stem Volume ◽  
Global Dynamics ◽  
Vegetation Density ◽  
Local Data ◽  
Horizontal Structure ◽  
Model Training

The structure of forests is important to observe for understanding coupling to global dynamics of ecosystems, biodiversity, and management aspects. In this paper, the sensitivity of X-band to boreal forest stem volume and to vertical and horizontal structure in the form of forest height and horizontal vegetation density is studied using TanDEM-X satellite observations from two study sites in Sweden: Remningstorp and Krycklan. The forest was analyzed with the Interferometric Water Cloud Model (IWCM), without the use of local data for model training, and compared with measurements by Airborne Lidar Scanning (ALS). On one hand, a large number of stands were studied, and in addition, plots with different types of changes between 2010 and 2014 were also studied. It is shown that the TanDEM-X phase height is, under certain conditions, equal to the product of the ALS quantities for height and density. Therefore, the sensitivity of phase height to relative changes in height and density is the same. For stands with a phase height >5 m we obtained an root-mean-square error, RMSE, of 8% and 10% for tree height in Remningstorp and Krycklan, respectively, and for vegetation density an RMSE of 13% for both. Furthermore, we obtained an RMSE of 17% for estimation of above ground biomass at stand level in Remningstorp and in Krycklan. The forest changes estimated with TanDEM-X/IWCM and ALS are small for all plots except clear cuts but show similar trends. Plots without forest management changes show a mean estimated height growth of 2.7% with TanDEM-X/IWCM versus 2.1% with ALS and a biomass growth of 4.3% versus 4.2% per year. The agreement between the estimates from TanDEM-X/IWCM and ALS is in general good, except for stands with low phase height.

Train++: An Incremental ML Model Training Algorithm to Create Self-Learning IoT Devices

2021 ◽  
Author(s):  
Bharath Sudharsan ◽  
Piyush Yadav ◽  
John G. Breslin ◽  
Muhammad Intizar Ali
Keyword(s):  
Training Algorithm ◽  
Model Training ◽  
Iot Devices ◽  
Self Learning

Globe2Train: A Framework for Distributed ML Model Training using IoT Devices Across the Globe

2021 ◽  
Author(s):  
Bharath Sudharsan ◽  
John G. Breslin ◽  
Muhammad Intizar Ali
Keyword(s):  
Model Training ◽  
Iot Devices

scholarly journals A Framework for Malicious Traffic Detection in IoT Healthcare Environment

Sensors ◽  
2021 ◽  
Vol 21 (9) ◽  
pp. 3025
Author(s):  
Faisal Hussain ◽  
Syed Ghazanfar Abbas ◽  
Ghalib A. Shah ◽  
Ivan Miguel Pires ◽  
Ubaid U. Fayyaz ◽  
...  
Keyword(s):  
Open Source ◽  
Human Life ◽  
Cyber Attacks ◽  
Smart Devices ◽  
Use Case ◽  
Context Aware ◽  
Iot Security ◽  
Healthcare Environment ◽  
Data Generator ◽  
Iot Devices

The Internet of things (IoT) has emerged as a topic of intense interest among the research and industrial community as it has had a revolutionary impact on human life. The rapid growth of IoT technology has revolutionized human life by inaugurating the concept of smart devices, smart healthcare, smart industry, smart city, smart grid, among others. IoT devices’ security has become a serious concern nowadays, especially for the healthcare domain, where recent attacks exposed damaging IoT security vulnerabilities. Traditional network security solutions are well established. However, due to the resource constraint property of IoT devices and the distinct behavior of IoT protocols, the existing security mechanisms cannot be deployed directly for securing the IoT devices and network from the cyber-attacks. To enhance the level of security for IoT, researchers need IoT-specific tools, methods, and datasets. To address the mentioned problem, we provide a framework for developing IoT context-aware security solutions to detect malicious traffic in IoT use cases. The proposed framework consists of a newly created, open-source IoT data generator tool named IoT-Flock. The IoT-Flock tool allows researchers to develop an IoT use-case comprised of both normal and malicious IoT devices and generate traffic. Additionally, the proposed framework provides an open-source utility for converting the captured traffic generated by IoT-Flock into an IoT dataset. Using the proposed framework in this research, we first generated an IoT healthcare dataset which comprises both normal and IoT attack traffic. Afterwards, we applied different machine learning techniques to the generated dataset to detect the cyber-attacks and protect the healthcare system from cyber-attacks. The proposed framework will help in developing the context-aware IoT security solutions, especially for a sensitive use case like IoT healthcare environment.

scholarly journals Performance Evaluation of Publish-Subscribe Systems in IoT Using Energy-Efficient and Context-Aware Secure Messages

2021 ◽  
Author(s):  
Norisvaldo Ferraz Junior ◽  
Anderson AA Silva ◽  
Adilson E Guelfi ◽  
Sergio T Kofuji
Keyword(s):  
Energy Efficiency ◽  
Performance Evaluation ◽  
Energy Efficient ◽  
High Capacity ◽  
Context Aware ◽  
Cloud Platform ◽  
Ultra Low Power ◽  
Message Structure ◽  
End To End ◽  
Iot Devices

Abstract Background: The Internet of Things (IoT) enables the development of innovative applications in various domains such as healthcare, transportation, and Industry 4.0. Publish-subscribe systems enable IoT devices to communicate with the cloud platform. However, IoT applications need context-aware messages to translate the data into contextual information, allowing the applications to act cognitively. Besides, end-to-end security of publish-subscribe messages on both ends (devices and cloud) is essential. However, achieving security on constrained IoT devices with memory, payload, and energy restrictions is a challenge. Contribution: Messages in IoT need to achieve both energy efficiency and secure delivery. Thus, the main contribution of this paper refers to a performance evaluation of a message structure that standardizes the publish-subscribe topic and payload used by the cloud platform and the IoT devices. We also propose a standardization for the topic and payload for publish-subscribe systems. Conclusion: The messages promote energy efficiency, enabling ultra-low-power and high-capacity devices and reducing the bytes transmitted in the IoT domain. The performance evaluation demonstrates that publish-subscribe systems (namely, AMQP, DDS, and MQTT) can use our proposed energy-efficient message structure on IoT. Additionally, the message system provides end-to-end confidentiality, integrity, and authenticity between IoT devices and the cloud platform.

Sign in / Sign up

Export Citation Format

Share Document