Biologically inspired adaptive neuro-filtering

The adaptive properties of sensory cells are very important to the reliable operation of biological vision systems. Sensory cells have evolved to make near optimum use of limited bandwidth channels and computational resources by extracting only the necessary information from the environment. This near optimal use of hardware has important implications for designers of artificial sensors and processing systems and it is therefore useful to understand the operation of visual cells. The paper proposes and analyses some simple filter designs that mimic some of the properties of large monopolar cells in the insect visual system. The filters use a well known neural mechanism, known as shunting inhibition, as their basic component.