Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review
Secure Machine Intelligence and Distributed Ledger. / Arseniev, Dmitry ; Baskakov, Dmitry; Shkodyrev, Vyacheslav.
Computational Science and Its Applications – ICCSA 2021: 21st International Conference, Cagliari, Italy, September 13–16, 2021, Proceedings, Part III. ed. / O Gervasi; B Murgante; S Misra; C Garau; Blecic; D Taniar; BO Apduhan; AMAC Rocha; E Tarantino; CM Torre. Cham : Springer Nature, 2021. p. 227-239 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12951 LNCS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review
}
TY - GEN
T1 - Secure Machine Intelligence and Distributed Ledger
AU - Arseniev, Dmitry
AU - Baskakov, Dmitry
AU - Shkodyrev, Vyacheslav
N1 - Arseniev D., Baskakov D., Shkodyrev V. (2021) Secure Machine Intelligence and Distributed Ledger. In: Gervasi O. et al. (eds) Computational Science and Its Applications – ICCSA 2021. ICCSA 2021. Lecture Notes in Computer Science, vol 12951. Springer, Cham. https://doi.org/10.1007/978-3-030-86970-0_17
PY - 2021
Y1 - 2021
N2 - Modern machine and deep learning systems are becoming part of high-performance cloud services and technologies. It is extremely important to understand that in systems such as recommendation systems, data is stored on local machines, and the trained system (matrix) is located in the cloud vendor, for example, in AWS or Google Cloud. Data on local machines can be updated or deleted. Local machines are often networked, which requires the use of synchronization methods and specialized protocols for the exchange of such information. And the central server is used as a single computer center for machine learning tasks. At the same time, it is necessary to control both the integrity of local data and their relevance with respect to other local machines. An important aspect is that the data center in the cloud should not know about our data, that is, we must be able to transmit them in encrypted form. At the same time, the deep learning model should be able to work with such encrypted data and send us the answers in encrypted form too. All this should be calculated in polynomial time, that is, quickly enough. For encryption purposes, it is proposed to use homomorphic algorithms. This report attempts to combine two promising modern paradigms for solving similar problems: machine intelligence and distributed ledger. For the purposes of distributed deep learning in relation to recommender systems, this symbiosis shows very serious practical prospects.
AB - Modern machine and deep learning systems are becoming part of high-performance cloud services and technologies. It is extremely important to understand that in systems such as recommendation systems, data is stored on local machines, and the trained system (matrix) is located in the cloud vendor, for example, in AWS or Google Cloud. Data on local machines can be updated or deleted. Local machines are often networked, which requires the use of synchronization methods and specialized protocols for the exchange of such information. And the central server is used as a single computer center for machine learning tasks. At the same time, it is necessary to control both the integrity of local data and their relevance with respect to other local machines. An important aspect is that the data center in the cloud should not know about our data, that is, we must be able to transmit them in encrypted form. At the same time, the deep learning model should be able to work with such encrypted data and send us the answers in encrypted form too. All this should be calculated in polynomial time, that is, quickly enough. For encryption purposes, it is proposed to use homomorphic algorithms. This report attempts to combine two promising modern paradigms for solving similar problems: machine intelligence and distributed ledger. For the purposes of distributed deep learning in relation to recommender systems, this symbiosis shows very serious practical prospects.
KW - Deep learning
KW - Differential privacy
KW - Homomorphic encryption
KW - Machine learning
KW - Secure computation
UR - http://www.scopus.com/inward/record.url?scp=85115716362&partnerID=8YFLogxK
UR - https://www.mendeley.com/catalogue/032f0340-1ae1-3156-9274-b08f13bd5d92/
U2 - 10.1007/978-3-030-86970-0_17
DO - 10.1007/978-3-030-86970-0_17
M3 - Conference contribution
AN - SCOPUS:85115716362
SN - 9783030869694
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 227
EP - 239
BT - Computational Science and Its Applications – ICCSA 2021
A2 - Gervasi, O
A2 - Murgante, B
A2 - Misra, S
A2 - Garau, C
A2 - Blecic, null
A2 - Taniar, D
A2 - Apduhan, BO
A2 - Rocha, AMAC
A2 - Tarantino, E
A2 - Torre, CM
PB - Springer Nature
CY - Cham
T2 - 21st International Conference on Computational Science and Its Applications, ICCSA 2021
Y2 - 13 September 2021 through 16 September 2021
ER -
ID: 86501406