Abstract: A system and method for creating at least one new network, comprising: selecting at least one node and at least one set of reactions where reagents in each reaction of the at least one set of reactions are known in at least one reference network; and creating at least one new network by causing the at least one new network to behave in a similar way with respect to the at least one node and the at least one set of reactions as the at least one reference network reacts with the at least one node and the at least one set of reactions.

Abstract: A denoising mechanism uses chosen signal classes and selected analysis dictionaries. The chosen signal class includes a collection of signals. The analysis dictionaries describe signals. The embedding threshold value is initially determined for a training set of signals in the chosen signal class. The update signal is initialized with a signal corrupted by noise.

Abstract: The invention provides for a technique of extracting information from signals by allowing a user to select a classifier and a figure of merit for quantifying classification quality; select a transform to generate features from input data; use a recursive process of functional dissipation to generate dissipative features from features that are generated according to the transform; search for one or more dissipative features that maximize the figure of merit on a training set; and classify a test set with the classifier by using one or more of the dissipative features. Functional dissipation uses the transforms recursively by generating random masking functions and extracting features with one or more generalized matching pursuit iterations. In each iteration, the recursive process may modify several features of the transformed signal with the largest absolute values according to a specific masking function.