Neural network learning device, method, and program

Abstract

The learning unitlearns a neural network. The linearization quantity determination unitdetermines linearization quantity, which is a parameter included in an activation function used in the neural network, and which is a parameter that brings the activation function closer to a linear function by increasing or decreasing itself. The aggregation unitreplaces the activation function, which is determined to converge to a linear function by increasing or decreasing the linearization quantity, with the linear function, and aggregating weights among layers using the replaced linear function. The learning unitcalculates evaluation value based on output by the neural network in learning the neural network, and the linearization quantity determination unitchanges the linearization quantity when the evaluation value satisfies the predetermined criterion.