2018
DOI: 10.1109/mc.2018.2381113
|View full text |Cite
|
Sign up to set email alerts
|

Private and Scalable Personal Data Analytics Using Hybrid Edge-to-Cloud Deep Learning

Abstract: We are observing an increasing presence of cyber-physical systems and their associated data around us. While the ability to collect, collate, and analyze the vast amount of rich information from smartphones, IoT devices, and urban sensors can be beneficial to the users and the industry, this process has led to a number of challenges ranging from performing efficient and meaningful analytics on the generated big data, to privacy challenges associated with the inferences made from these data due to ubiquitous na… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
22
0

Year Published

2019
2019
2023
2023

Publication Types

Select...
7
2
1

Relationship

2
8

Authors

Journals

citations
Cited by 39 publications
(22 citation statements)
references
References 11 publications
0
22
0
Order By: Relevance
“…As for efficiency, the obfuscation introduces a time overhead proportional to the input size, considered by the authors as acceptable for real-time processing. Osia et al [39,67,68] described a hybrid architecture for inference based on model splitting concept. The layers are split between local device and the cloud, and the process into feature extraction, performed by the local primary layers, and analysis, performed by secondary layers on the cloud.…”
Section: The Studied Scenarios Comprise Attacks : On Communication Bementioning
confidence: 99%
“…As for efficiency, the obfuscation introduces a time overhead proportional to the input size, considered by the authors as acceptable for real-time processing. Osia et al [39,67,68] described a hybrid architecture for inference based on model splitting concept. The layers are split between local device and the cloud, and the process into feature extraction, performed by the local primary layers, and analysis, performed by secondary layers on the cloud.…”
Section: The Studied Scenarios Comprise Attacks : On Communication Bementioning
confidence: 99%
“…In [51], an overview of algorithmic and processor techniques for transitioning deep learning to IoT and CPS is provided. A hybrid framework for privacy-preserving, accurate, and efficient analytics of IoT and CPS data based on user-centered edge devices and cloud computing is proposed in [52]. In [20], authors attempt to use industrial IoT to detect anomalies in power consumption based on a multi-view stacking intelligent ensemble.…”
Section: Internet Of Things (Iot) and Cyber-physical Systems (Cps)mentioning
confidence: 99%
“…In particular, we show how to generate natural-looking adversarial images either by selectively modifying colors within chosen ranges that we perceive as natural or by enhancing details in the image. The references that will be covered in the tutorial are [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18].…”
Section: Context Motivation and Descriptionmentioning
confidence: 99%