Synaptic Pruning Mechanisms in Learning

1 –Independent Work Report, Spring 2015– Synaptic Pruning Mechanisms in Learning Abstract Synaptic pruning is the process of removing synapses in neural networks and has been considered to be a method of learning. During the developmental stages of the brain, synaptic pruning helps regulate efficiency and energy conservation. However, a destructive algorithm seems to be counter-intuitive to “learning.” When applied to association networks, pruning networks show evidence of learning associations when given good input. However, the pruning algorithm can be detrimental to network size and even limit learning for networks that are too small or have too high of a frequency threshold. For small networks, implementing constructive algorithm in addition to the pruning one may help ensure that a network continues to grow and learn.