Focusing on improving the dead units problem of the
classical CL, there are also several other techniques proposed,
e.g., leaky learning by Rumelhart [7] and Grossberg [8], and
convex bridge by Hecht-Nielsen [9]. A notable improvement
is the strategy of reducing the winning rate of the frequent
winners [8], [171-[ 193, sometimes called conscience [181.
Frequency Sensitive Competitive Learning (FSCL) [19] is a
good example which uses this strategy. The idea is fairly
straightforward but the method does improve the classical CL
significantly