Regulatory networks studied by ellipsoidal calculus

Download
2015
Yayla, Selim
The identification of regulatory networks affected by noise and data uncertainty is a serious problem in many Operational Research applications. The fundamental structure of underlying systems can be established by regulatory networks in many sector like ecology, education and finance. After clustering and classification methods gene/target and environmental states can be grouped into functional behaviour. The analysis of complex regulatory systems under uncertainty is a compounded complex by the unknown interactions between the variables which are represented by ellipsoids. Ellipsoidal calculus is used in determination of the explicit representations of the uncertain multivariate states of the system. MATLAB Ellipsoidal Toolbox (ET) provides efficient plotting routines of ellipsoids, hyperplanes and reach sets. In this thesis, several regression models are studied in order to approximate regulatory networks under ellipsoidal uncertainty and Ellipsoidal Toolbox routines are explained for representing a parameter estimation and inverse problem.