Evidence Evaluation: a Study of Likelihoods and Independence

Silja Renooij
Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:426-437, 2016.

Abstract

In the context of evidence evaluation, where the probability of evidence given a certain hypothesis is considered, different pieces of evidence are often combined in a naive way by assuming conditional independence. In this paper we present a number of results that can be used to assess both the importance of a reliable likelihood-ratio estimate and the impact of neglecting dependencies among pieces of evidence for the purpose of evidence evaluation. We analytically study the effect of changes in dependencies between pieces of evidence on the likelihood ratio, and provide both theoretical and empirical bounds on the error in likelihood occasioned by assuming independences that do not hold in practice. In addition, a simple measure of influence strength between pieces of evidence is proposed.

Cite this Paper

BibTeX
@InProceedings{pmlr-v52-renooij16, title = {Evidence Evaluation: a Study of Likelihoods and Independence}, author = {Renooij, Silja}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {426--437}, year = {2016}, editor = {Antonucci, Alessandro and Corani, Giorgio and Campos}, Cassio Polpo}, volume = {52}, series = {Proceedings of Machine Learning Research}, address = {Lugano, Switzerland}, month = {06--09 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v52/renooij16.pdf}, url = {https://proceedings.mlr.press/v52/renooij16.html}, abstract = {In the context of evidence evaluation, where the probability of evidence given a certain hypothesis is considered, different pieces of evidence are often combined in a naive way by assuming conditional independence. In this paper we present a number of results that can be used to assess both the importance of a reliable likelihood-ratio estimate and the impact of neglecting dependencies among pieces of evidence for the purpose of evidence evaluation. We analytically study the effect of changes in dependencies between pieces of evidence on the likelihood ratio, and provide both theoretical and empirical bounds on the error in likelihood occasioned by assuming independences that do not hold in practice. In addition, a simple measure of influence strength between pieces of evidence is proposed.} }
Endnote
%0 Conference Paper %T Evidence Evaluation: a Study of Likelihoods and Independence %A Silja Renooij %B Proceedings of the Eighth International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2016 %E Alessandro Antonucci %E Giorgio Corani %E Cassio Polpo Campos} %F pmlr-v52-renooij16 %I PMLR %P 426--437 %U https://proceedings.mlr.press/v52/renooij16.html %V 52 %X In the context of evidence evaluation, where the probability of evidence given a certain hypothesis is considered, different pieces of evidence are often combined in a naive way by assuming conditional independence. In this paper we present a number of results that can be used to assess both the importance of a reliable likelihood-ratio estimate and the impact of neglecting dependencies among pieces of evidence for the purpose of evidence evaluation. We analytically study the effect of changes in dependencies between pieces of evidence on the likelihood ratio, and provide both theoretical and empirical bounds on the error in likelihood occasioned by assuming independences that do not hold in practice. In addition, a simple measure of influence strength between pieces of evidence is proposed.
RIS
TY - CPAPER TI - Evidence Evaluation: a Study of Likelihoods and Independence AU - Silja Renooij BT - Proceedings of the Eighth International Conference on Probabilistic Graphical Models DA - 2016/08/15 ED - Alessandro Antonucci ED - Giorgio Corani ED - Cassio Polpo Campos} ID - pmlr-v52-renooij16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 52 SP - 426 EP - 437 L1 - http://proceedings.mlr.press/v52/renooij16.pdf UR - https://proceedings.mlr.press/v52/renooij16.html AB - In the context of evidence evaluation, where the probability of evidence given a certain hypothesis is considered, different pieces of evidence are often combined in a naive way by assuming conditional independence. In this paper we present a number of results that can be used to assess both the importance of a reliable likelihood-ratio estimate and the impact of neglecting dependencies among pieces of evidence for the purpose of evidence evaluation. We analytically study the effect of changes in dependencies between pieces of evidence on the likelihood ratio, and provide both theoretical and empirical bounds on the error in likelihood occasioned by assuming independences that do not hold in practice. In addition, a simple measure of influence strength between pieces of evidence is proposed. ER -
APA
Renooij, S.. (2016). Evidence Evaluation: a Study of Likelihoods and Independence. Proceedings of the Eighth International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 52:426-437 Available from https://proceedings.mlr.press/v52/renooij16.html.

Related Material