Cooperative training for attribute-distributed data: Trade-off between data transmission and performance

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Scopus citations

Abstract

This paper introduces a modeling framework for distributed regression with agents/experts observing attribute-distributed data (heterogeneous data). Under this model, a new algorithm, the iterative covariance optimization algorithm (ICOA), is designed to reshape the covariance matrix of the training residuals of individual agents so that the linear combination of the individual estimators minimizes the ensemble training error. Moreover, a scheme (Minimax Protection) is designed to provide a trade-off between the number of data instances transmitted among the agents and the performance of the ensemble estimator without undermining the convergence of the algorithm. This scheme also provides an upper bound (with high probability) on the test error of the ensemble estimator. The efficacy of ICOA combined with Minimax Protection and the comparison between the upper bound and actual performance are both demonstrated by simulations.

Original languageEnglish (US)
Title of host publication2009 12th International Conference on Information Fusion, FUSION 2009
Pages664-671
Number of pages8
StatePublished - Nov 18 2009
Event2009 12th International Conference on Information Fusion, FUSION 2009 - Seattle, WA, United States
Duration: Jul 6 2009Jul 9 2009

Publication series

Name2009 12th International Conference on Information Fusion, FUSION 2009

Other

Other2009 12th International Conference on Information Fusion, FUSION 2009
CountryUnited States
CitySeattle, WA
Period7/6/097/9/09

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Networks and Communications
  • Information Systems
  • Software

Keywords

  • Cooperative training
  • Distributed learning
  • Heterogeneous data

Fingerprint Dive into the research topics of 'Cooperative training for attribute-distributed data: Trade-off between data transmission and performance'. Together they form a unique fingerprint.

  • Cite this

    Zheng, H., Kulkarni, S. R., & Poor, H. V. (2009). Cooperative training for attribute-distributed data: Trade-off between data transmission and performance. In 2009 12th International Conference on Information Fusion, FUSION 2009 (pp. 664-671). [5203804] (2009 12th International Conference on Information Fusion, FUSION 2009).