Error correction and partial occlusion
Bio-inspired perception processor. A neuromorphic approach

An element corresponds to a set of elementary semantic representations, which may be incomplete and/or contain errors that need to be corrected. An automaton integrated into the associative memory will select the greatest number of elementary semantic representations common to an element, from the first sub-module, in order to provide the right link to the second module, which will output the corresponding word. In the example shown in figure