Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Pruning (artificial neural network)
open-in-new
<p>In <a href="/facts/Deep_learning/JLuwD3ea">deep learning</a>, pruning is the practice of removing <a href="/facts/Parameter/zFvVFMv9">parameters</a> from an existing <a href="/facts/Neural_network_(machine_learning)/6V1jMlkx">artificial neural network</a>. The goal of this process is to reduce the size (parameter count) of the neural network (and therefore the <a href="/facts/Computational_resource/YyleNx71">computational resources</a> required to run it) whilst maintaining accuracy. This can be compared to the biological process of <a href="/facts/Synaptic_pruning/H39FHLqe">synaptic pruning</a> which takes place in <a href="/facts/Mammal/phSwTzBo">mammalian</a> brains during development. </p>