Max-Min-Degree Neural Network for Centralized-Decentralized Collaborative Computing

Yiqiang SHENG; Jinlin WANG; Chaopeng LI; Weining QI

doi:10.1587/transcom.2015ADP0013

Max-Min-Degree Neural Network for Centralized-Decentralized Collaborative Computing

Yiqiang SHENG, Jinlin WANG, Chaopeng LI, Weining QI

Full Text Views

0

Share
Cite this

Summary :

In this paper, we propose an undirected model of learning systems, named max-min-degree neural network, to realize centralized-decentralized collaborative computing. The basic idea of the proposal is a max-min-degree constraint which extends a k-degree constraint to improve the communication cost, where k is a user-defined degree of neurons. The max-min-degree constraint is defined such that the degree of each neuron lies between k_min and k_max. Accordingly, the Boltzmann machine is a special case of the proposal with k_min=k_max=n, where n is the full-connected degree of neurons. Evaluations show that the proposal is much better than a state-of-the-art model of deep learning systems with respect to the communication cost. The cost of the above improvement is slower convergent speed with respect to data size, but it does not matter in the case of big data processing.

Publication: IEICE TRANSACTIONS on Communications Vol.E99-B No.4 pp.841-848

Publication Date: 2016/04/01

Publicized

Online ISSN: 1745-1345

DOI: 10.1587/transcom.2015ADP0013

Type of Manuscript: Special Section PAPER (Special Section on Autonomous Decentralized Systems Technologies and Applications for Next-Generation Social Infrastructure)

Category

Authors

Yiqiang SHENG
  Chinese Academy of Sciences
Jinlin WANG
  Chinese Academy of Sciences
Chaopeng LI
  University of Chinese Academy of Sciences
Weining QI
  Chinese Academy of Sciences

Keyword

big data, cloud computing, decentralized computing, collaborative computing, learning systems

Cite this

Copy

Yiqiang SHENG, Jinlin WANG, Chaopeng LI, Weining QI, "Max-Min-Degree Neural Network for Centralized-Decentralized Collaborative Computing" in IEICE TRANSACTIONS on Communications, vol. E99-B, no. 4, pp. 841-848, April 2016, doi: 10.1587/transcom.2015ADP0013.
Abstract: In this paper, we propose an undirected model of learning systems, named max-min-degree neural network, to realize centralized-decentralized collaborative computing. The basic idea of the proposal is a max-min-degree constraint which extends a k-degree constraint to improve the communication cost, where k is a user-defined degree of neurons. The max-min-degree constraint is defined such that the degree of each neuron lies between k_min and k_max. Accordingly, the Boltzmann machine is a special case of the proposal with k_min=k_max=n, where n is the full-connected degree of neurons. Evaluations show that the proposal is much better than a state-of-the-art model of deep learning systems with respect to the communication cost. The cost of the above improvement is slower convergent speed with respect to data size, but it does not matter in the case of big data processing.
URL: https://globals.ieice.org/en_transactions/communications/10.1587/transcom.2015ADP0013/_p

Copy

@ARTICLE{e99-b_4_841,
author={Yiqiang SHENG, Jinlin WANG, Chaopeng LI, Weining QI, },
journal={IEICE TRANSACTIONS on Communications},
title={Max-Min-Degree Neural Network for Centralized-Decentralized Collaborative Computing},
year={2016},
volume={E99-B},
number={4},
pages={841-848},
abstract={In this paper, we propose an undirected model of learning systems, named max-min-degree neural network, to realize centralized-decentralized collaborative computing. The basic idea of the proposal is a max-min-degree constraint which extends a k-degree constraint to improve the communication cost, where k is a user-defined degree of neurons. The max-min-degree constraint is defined such that the degree of each neuron lies between k_min and k_max. Accordingly, the Boltzmann machine is a special case of the proposal with k_min=k_max=n, where n is the full-connected degree of neurons. Evaluations show that the proposal is much better than a state-of-the-art model of deep learning systems with respect to the communication cost. The cost of the above improvement is slower convergent speed with respect to data size, but it does not matter in the case of big data processing.},
keywords={},
doi={10.1587/transcom.2015ADP0013},
ISSN={1745-1345},
month={April},}

Copy

TY - JOUR
TI - Max-Min-Degree Neural Network for Centralized-Decentralized Collaborative Computing
T2 - IEICE TRANSACTIONS on Communications
SP - 841
EP - 848
AU - Yiqiang SHENG
AU - Jinlin WANG
AU - Chaopeng LI
AU - Weining QI
PY - 2016
DO - 10.1587/transcom.2015ADP0013
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E99-B
IS - 4
JA - IEICE TRANSACTIONS on Communications
Y1 - April 2016
AB - In this paper, we propose an undirected model of learning systems, named max-min-degree neural network, to realize centralized-decentralized collaborative computing. The basic idea of the proposal is a max-min-degree constraint which extends a k-degree constraint to improve the communication cost, where k is a user-defined degree of neurons. The max-min-degree constraint is defined such that the degree of each neuron lies between k_min and k_max. Accordingly, the Boltzmann machine is a special case of the proposal with k_min=k_max=n, where n is the full-connected degree of neurons. Evaluations show that the proposal is much better than a state-of-the-art model of deep learning systems with respect to the communication cost. The cost of the above improvement is slower convergent speed with respect to data size, but it does not matter in the case of big data processing.
ER -