Neural network pruning is conducive to the simplification of network structure. However, as a relatively important pruning algorithm, the correlation pruning algorithm does not give a clear boundary for how to delete nodes according to the linear correlation value and variance value after calculating the linear correlation and variance of the hidden layer node output. This paper studies the correlation pruning algorithm of neural network and gives a method of deleting nodes according to the variance value based on the error propagation of the network. Experiments show that this method can not only effectively simplify the network structure and ensure the network accuracy, but also is simple to calculate.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore