Deep neural network-based method and device for quantifying activation amount

Abstract

A method and an apparatus for quantizing an activation volume of a deep neural network are disclosed. The method includes: obtaining an activation volume of a network layer in the deep neural network (S), wherein elements in the activation volume are arranged in three directions: height, width, and depth; dividing depth segments in the activation volume in which a difference among element features is smaller than a preset threshold into a same slice group along the depth direction of the activation volume, so as to obtain a plurality of slice groups (S); quantizing each slice group respectively by using a quantization parameter corresponding to each slice group obtained through a quantization formula (S). The quantization error can be reduced through the above method.