Inferring latent states in a network influenced by neighbor activities: An undirected generative approach

Buddhika L. Samarakoon, Manohar Murthi, Kamal Premaratne

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

Abstract

The problem of inferring the hidden state of individual nodes in social/sensor networks in which node activities affect their neighbors is growing in importance. We present an undirected generative model, a type of probabilistic model that has so far not been used for modeling latent variables influenced by neighbors in a network. We also propose an efficient inference method based on variational inference principles which, in contrast to sampling methods used in most existing models, is scalable to larger networks. While training is intractable in general, by using stochastic methods to approximate the intractable derivative, we show that our model can be trained using the maximum likelihood method by formulating the model as an exponential family distribution. The results demonstrate that the proposed undirected model can accurately infer latent states compared to baseline methods.

Original languageEnglish (US)
Title of host publication2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages2372-2376
Number of pages5
ISBN (Electronic)9781509041176
DOIs
StatePublished - Jun 16 2017
Event2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - New Orleans, United States
Duration: Mar 5 2017Mar 9 2017

Other

Other2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017
CountryUnited States
CityNew Orleans
Period3/5/173/9/17

Fingerprint

Maximum likelihood
Sensor networks
Sampling
Derivatives
Statistical Models

Keywords

  • exponential family
  • factor models
  • latent sentiments
  • Network influence
  • stochastic gradient methods

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Samarakoon, B. L., Murthi, M., & Premaratne, K. (2017). Inferring latent states in a network influenced by neighbor activities: An undirected generative approach. In 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings (pp. 2372-2376). [7952581] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2017.7952581

Inferring latent states in a network influenced by neighbor activities : An undirected generative approach. / Samarakoon, Buddhika L.; Murthi, Manohar; Premaratne, Kamal.

2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. p. 2372-2376 7952581.

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

Samarakoon, BL, Murthi, M & Premaratne, K 2017, Inferring latent states in a network influenced by neighbor activities: An undirected generative approach. in 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings., 7952581, Institute of Electrical and Electronics Engineers Inc., pp. 2372-2376, 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017, New Orleans, United States, 3/5/17. https://doi.org/10.1109/ICASSP.2017.7952581
Samarakoon BL, Murthi M, Premaratne K. Inferring latent states in a network influenced by neighbor activities: An undirected generative approach. In 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2017. p. 2372-2376. 7952581 https://doi.org/10.1109/ICASSP.2017.7952581
Samarakoon, Buddhika L. ; Murthi, Manohar ; Premaratne, Kamal. / Inferring latent states in a network influenced by neighbor activities : An undirected generative approach. 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 2372-2376
@inproceedings{07a5b893df6e44339b8298dcea9cb1c7,
title = "Inferring latent states in a network influenced by neighbor activities: An undirected generative approach",
abstract = "The problem of inferring the hidden state of individual nodes in social/sensor networks in which node activities affect their neighbors is growing in importance. We present an undirected generative model, a type of probabilistic model that has so far not been used for modeling latent variables influenced by neighbors in a network. We also propose an efficient inference method based on variational inference principles which, in contrast to sampling methods used in most existing models, is scalable to larger networks. While training is intractable in general, by using stochastic methods to approximate the intractable derivative, we show that our model can be trained using the maximum likelihood method by formulating the model as an exponential family distribution. The results demonstrate that the proposed undirected model can accurately infer latent states compared to baseline methods.",
keywords = "exponential family, factor models, latent sentiments, Network influence, stochastic gradient methods",
author = "Samarakoon, {Buddhika L.} and Manohar Murthi and Kamal Premaratne",
year = "2017",
month = "6",
day = "16",
doi = "10.1109/ICASSP.2017.7952581",
language = "English (US)",
pages = "2372--2376",
booktitle = "2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Inferring latent states in a network influenced by neighbor activities

T2 - An undirected generative approach

AU - Samarakoon, Buddhika L.

AU - Murthi, Manohar

AU - Premaratne, Kamal

PY - 2017/6/16

Y1 - 2017/6/16

N2 - The problem of inferring the hidden state of individual nodes in social/sensor networks in which node activities affect their neighbors is growing in importance. We present an undirected generative model, a type of probabilistic model that has so far not been used for modeling latent variables influenced by neighbors in a network. We also propose an efficient inference method based on variational inference principles which, in contrast to sampling methods used in most existing models, is scalable to larger networks. While training is intractable in general, by using stochastic methods to approximate the intractable derivative, we show that our model can be trained using the maximum likelihood method by formulating the model as an exponential family distribution. The results demonstrate that the proposed undirected model can accurately infer latent states compared to baseline methods.

AB - The problem of inferring the hidden state of individual nodes in social/sensor networks in which node activities affect their neighbors is growing in importance. We present an undirected generative model, a type of probabilistic model that has so far not been used for modeling latent variables influenced by neighbors in a network. We also propose an efficient inference method based on variational inference principles which, in contrast to sampling methods used in most existing models, is scalable to larger networks. While training is intractable in general, by using stochastic methods to approximate the intractable derivative, we show that our model can be trained using the maximum likelihood method by formulating the model as an exponential family distribution. The results demonstrate that the proposed undirected model can accurately infer latent states compared to baseline methods.

KW - exponential family

KW - factor models

KW - latent sentiments

KW - Network influence

KW - stochastic gradient methods

UR - http://www.scopus.com/inward/record.url?scp=85023768616&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85023768616&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2017.7952581

DO - 10.1109/ICASSP.2017.7952581

M3 - Conference contribution

AN - SCOPUS:85023768616

SP - 2372

EP - 2376

BT - 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -