Online EM-Based Ensemble Classification With Correlated Agents

A binary ensemble classification method that sequentially processes the data collected from multiple decision agents in the presence of parameter uncertainties is proposed. Agents are assumed to form correlated groups whose decisions are modeled as multivariate Bernoulli random vectors. The prior probabilities of the binary hypotheses and the corresponding probabilities of the outcomes under each hypothesis are treated as unknown deterministic parameters. Cappé's online Expectation-Maximization algorithm is employed to estimate the parameter values, which are then fed into the ensemble classifier. The proposed technique is shown the reduce the computational complexity while delivering performance close to its offline counterpart, which requires multiple passes over the data. Numerical examples are presented to corroborate the results.

[1]  Burak Kantarci,et al.  Detecting Fake Mobile Crowdsensing Tasks: Ensemble Methods Under Limited Data , 2020, IEEE Vehicular Technology Magazine.

[2]  Jiang-She Zhang,et al.  Robust CP Tensor Factorization With Skew Noise , 2020, IEEE Signal Processing Letters.

[3]  Gerardo Hermosillo,et al.  Learning From Crowds , 2010, J. Mach. Learn. Res..

[4]  Pranjal Awasthi,et al.  Crowdsourcing with Arbitrary Adversaries , 2018, ICML.

[5]  Yuval Kluger,et al.  Estimating the accuracies of multiple classifiers without labeled data , 2014, AISTATS.

[6]  Kaizhu Huang,et al.  Combination of Classification and Clustering Results with Label Propagation , 2014, IEEE Signal Processing Letters.

[7]  G. Wahba,et al.  Multivariate Bernoulli distribution , 2012, 1206.1874.

[8]  Roberto López-Valcarce,et al.  A Diffusion-Based EM Algorithm for Distributed Estimation in Unreliable Sensor Networks , 2013, IEEE Signal Processing Letters.

[9]  Georgios B. Giannakis,et al.  Blind Multiclass Ensemble Classification , 2017, IEEE Transactions on Signal Processing.

[10]  Xi Chen,et al.  Spectral Methods Meet EM: A Provably Optimal Algorithm for Crowdsourcing , 2014, J. Mach. Learn. Res..

[11]  Yuval Kluger,et al.  Ranking and combining multiple predictors without labeled data , 2013, Proceedings of the National Academy of Sciences.

[12]  Alba Pagès-Zamora,et al.  Unsupervised Ensemble Classification With Correlated Decision Agents , 2019, IEEE Signal Processing Letters.

[13]  Lior Rokach,et al.  Ensemble learning: A survey , 2018, WIREs Data Mining Knowl. Discov..

[14]  O. Cappé,et al.  On‐line expectation–maximization algorithm for latent data models , 2009 .

[15]  Jisheng Dai,et al.  Root Sparse Bayesian Learning for Off-Grid DOA Estimation , 2016, IEEE Signal Processing Letters.

[16]  David J. Miller,et al.  Critic-driven ensemble classification , 1999, IEEE Trans. Signal Process..

[17]  A. P. Dawid,et al.  Maximum Likelihood Estimation of Observer Error‐Rates Using the EM Algorithm , 1979 .

[18]  Xianfeng Zhao,et al.  Boosting Image Steganalysis Under Universal Deep Learning Architecture Incorporating Ensemble Classification Strategy , 2019, IEEE Signal Processing Letters.

[19]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[20]  Yuval Kluger,et al.  Unsupervised Ensemble Learning with Dependent Classifiers , 2015, AISTATS.

[21]  Antonio Pescapè,et al.  Multi-classification approaches for classifying mobile app traffic , 2018, J. Netw. Comput. Appl..

[22]  Pierluigi Salvo Rossi,et al.  A Systematic Framework for Composite Hypothesis Testing of Independent Bernoulli Trials , 2015, IEEE Signal Processing Letters.