Comparison of two inference approaches in Gaussian graphical models

Purutçuoğlu, V., Ayyıldız, E. & Wit, E. May-2017 In : Turkish journal of biochemistry-Turk biyokimya dergisi. 42, 2, p. 203-211 9 p.

Research output: Scientific - peer-reviewArticle

Introduction: The Gaussian Graphical Model (GGM) is one of the well-known probabilistic models which is based on the conditional independency of nodes in the biological system. Here, we compare the estimates of the GGM parameters by the graphical lasso (glasso) method and the threshold gradient descent (TGD) algorithm. Methods: We evaluate the performance of both techniques via certain measures such as specificity, F-measure and AUC (area under the curve). The analyses are conducted by Monte Carlo runs under different dimensional systems. Results: The results indicate that the TGD algorithm is more accurate than the glasso method in all selected criteria, whereas, it is more computationally demanding than this method too. Discussion and conclusion: Therefore, in high dimensional systems, we recommend glasso for its computational efficiency in spite of its loss in accuracy and we believe than the computational cost of the TGD algorithm can be improved by suggesting alternative steps in inference of the network.
Original languageEnglish
Pages (from-to)203-211
Number of pages9
JournalTurkish journal of biochemistry-Turk biyokimya dergisi
Issue number2
StatePublished - May-2017


  • Biological system, Gaussian graphical model, Graphical lasso, Model performance criteria, Threshold gradient descent algorithm, area under the curve, biology, controlled study, model, OIL, VARIABLE SELECTION, ORACLE PROPERTIES, LASSO, PATHWAY, LIKELIHOOD, REGRESSION, BIOLOGY

View graph of relations

ID: 47544773