Complex and emergent behavior from neural network pulse stream filters

Simple neural systems based on biologically motivated processing elements with applications to pulse stream filtering are discussed. These few node, moderately connected systems exhibit sensitive filtering capabilities, and complex and emergent information encoding properties. A single filter may operate on frequencies from the sub-hertz range up to the gigahertz range by simply adjusting the DC biases of the system. No changes in the physical structure of the system are required. Circuit level simulations of the systems are used to verify system functionality and demonstrate the basic waveforms produced by the hybrid processing element