Signal Processing by an Immune Type Tree Transform

Immunocomputing for intelligent signal processing

“…Then the reduced set of points after apoptosis and immunization can represent the feature extraction by FIN and the quality of such FIN can be estimated by the index of inseparability and thus can be compared with other FINs (e.g., those obtained by a preprocessing of the signal). On the other hand, let us note once again that the SVD can model the binding energy between proteins [3], whereas the dyadic DTT can model an immune-type antigen processing [29,31].…”

“…According to [29][30][31], the proposed approach to signal processing is inspired by a mode of biomolecular computing [15] when immune cells chop unknown antigen to its local singularities and expose them to the immune system. Analogously, the IC approach represents unknown signal as a tree of data, and chop the branches of the tree at the level l to detect local singularities of the signal.…”

“…A connection of IC with cellular automata (CA) [23] looks encouraging for threedimensional (3D) computer graphics [24] and identification of CA [25]. The IC is also connected with advances in applied self-organizing systems [26] including intelligent reconstruction and simulation of hydro-physical fields [27,28], signal processing [29][30][31] and intrusion detection [32][33][34].…”

“…The key model of the approach is the FIN. In the training mode, FIN can be formed from signal(s) using discrete tree transform (DTT) [29] and/or singular value decomposition (SVD) [35]. Such pre-processing by DTT and SVD provides the mapping of high-dimensional fragment of signal or combination of signals to the points (cells) of low-dimensional space of FIN.…”

An Ecological Approach to Anomaly Detection: The EIA Model

“…FIN proposes the generation of a Euclidean multidimensional space (FIN space) in which a training process based on application of a discrete tree transform (DTT) [1] and/or a singular value descomposition (SVD) [13] incorporates input data and its initial space, which is optimized to form a set of class representatives for representation called cytokines, following a process of apoptosis and immunization explained in [24]. These cytokines operate according to a proximity principle in the FIN space to determine the class of data reviewed when mapped in the space with the DTT algorithm.…”