On-line training and pruning for recursive least square algorithms

摘要

The authors derive a pruning method for recursive least square (RLS) based algorithms. In this approach, training (on-line type) and pruning are combined. Simulation results showed that this approach is an effective training and pruning method for neural networks.