Figure 1: (a) Actual circuit of the ts-WTA learning cell and (b) its abstract model. In (a) and and in (b) and show the floating gate based weighted connections. and are inputs and node voltage is activation of the cell which is equivalent to in (b). (c) shows ts-WTA evolution of floating gate voltages. (d) Starting with nearly equal weak connections (left), the cell strengthens the stronger of the two connections at the cost of the other (right, shows both possibilities). Here ○ implies connection representing one feature and ● implies connection representing other features (adapted from Gupta and Markan, 2014, [15]).