Limits...
MuTE: a MATLAB toolbox to compare established and novel estimators of the multivariate transfer entropy.

Montalto A, Faes L, Marinazzo D - PLoS ONE (2014)

Bottom Line: It is then safer that the method of choice for analyzing these interactions does not rely on any model or assumption on the nature of the data and their interactions.Transfer entropy has emerged as a powerful tool to quantify directed dynamical interactions.Applications to simulated and real data are presented.

View Article: PubMed Central - PubMed

Affiliation: Data Analysis Department, Ghent University, Ghent, Belgium.

ABSTRACT
A challenge for physiologists and neuroscientists is to map information transfer between components of the systems that they study at different scales, in order to derive important knowledge on structure and function from the analysis of the recorded dynamics. The components of physiological networks often interact in a nonlinear way and through mechanisms which are in general not completely known. It is then safer that the method of choice for analyzing these interactions does not rely on any model or assumption on the nature of the data and their interactions. Transfer entropy has emerged as a powerful tool to quantify directed dynamical interactions. In this paper we compare different approaches to evaluate transfer entropy, some of them already proposed, some novel, and present their implementation in a freeware MATLAB toolbox. Applications to simulated and real data are presented.

Show MeSH
ROC curves for all methods for the non-linear system.The curves are obtained reporting the results obtained gradually increasing the time series length simulated according to 14 from 128 to 1024 points.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4196918&req=5

pone-0109462-g008: ROC curves for all methods for the non-linear system.The curves are obtained reporting the results obtained gradually increasing the time series length simulated according to 14 from 128 to 1024 points.

Mentions: To understand how stable the performance of the methods is, in terms of sensitivity and the specificity, with respect to the length of the analyzed data set, we computed the analysis varying the series length from 128 to 1024 points. Figures 7 and 8 depict, respectively for the systems 13 and 14, the Receiver Operating Characteristic (ROC) curves obtained for all methods as a function of the series length. Evaluating the amount of TP (true positives), TN (true negatives), FP (false positives) and FN (false negatives) after grouping together all coupled directions (positives) and all uncoupled directions (negatives), we computed sensitivity as and specificity as . In the case of the linear system (Figure 7), all methods except the BIN UE provide good performance, with the LIN estimator providing the best sensitivity and specificity. All methods provided robust results with respect to the series length, with only a limited decay in the performance observed for 128 points. In the case of the non-linear system (Figure 8), the performance was optimal for BIN NUE and NN NUE (with a slightly lower specificity), while the methods implementing either the LIN estimator or the UE approach were considerably less sensitive.


MuTE: a MATLAB toolbox to compare established and novel estimators of the multivariate transfer entropy.

Montalto A, Faes L, Marinazzo D - PLoS ONE (2014)

ROC curves for all methods for the non-linear system.The curves are obtained reporting the results obtained gradually increasing the time series length simulated according to 14 from 128 to 1024 points.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4196918&req=5

pone-0109462-g008: ROC curves for all methods for the non-linear system.The curves are obtained reporting the results obtained gradually increasing the time series length simulated according to 14 from 128 to 1024 points.
Mentions: To understand how stable the performance of the methods is, in terms of sensitivity and the specificity, with respect to the length of the analyzed data set, we computed the analysis varying the series length from 128 to 1024 points. Figures 7 and 8 depict, respectively for the systems 13 and 14, the Receiver Operating Characteristic (ROC) curves obtained for all methods as a function of the series length. Evaluating the amount of TP (true positives), TN (true negatives), FP (false positives) and FN (false negatives) after grouping together all coupled directions (positives) and all uncoupled directions (negatives), we computed sensitivity as and specificity as . In the case of the linear system (Figure 7), all methods except the BIN UE provide good performance, with the LIN estimator providing the best sensitivity and specificity. All methods provided robust results with respect to the series length, with only a limited decay in the performance observed for 128 points. In the case of the non-linear system (Figure 8), the performance was optimal for BIN NUE and NN NUE (with a slightly lower specificity), while the methods implementing either the LIN estimator or the UE approach were considerably less sensitive.

Bottom Line: It is then safer that the method of choice for analyzing these interactions does not rely on any model or assumption on the nature of the data and their interactions.Transfer entropy has emerged as a powerful tool to quantify directed dynamical interactions.Applications to simulated and real data are presented.

View Article: PubMed Central - PubMed

Affiliation: Data Analysis Department, Ghent University, Ghent, Belgium.

ABSTRACT
A challenge for physiologists and neuroscientists is to map information transfer between components of the systems that they study at different scales, in order to derive important knowledge on structure and function from the analysis of the recorded dynamics. The components of physiological networks often interact in a nonlinear way and through mechanisms which are in general not completely known. It is then safer that the method of choice for analyzing these interactions does not rely on any model or assumption on the nature of the data and their interactions. Transfer entropy has emerged as a powerful tool to quantify directed dynamical interactions. In this paper we compare different approaches to evaluate transfer entropy, some of them already proposed, some novel, and present their implementation in a freeware MATLAB toolbox. Applications to simulated and real data are presented.

Show MeSH