A configurable neural network activation function implementation device

TL;DR: In this article, a configurable neural network activation function implementation device is presented, which consists of a controller, a symbol judgment module, a range detection module, an address generator, a parameter register, a floating point multiplier and a phase operand arithmetic unit.

Abstract: The invention discloses a configurable neural network activation function implementation device. The device comprises a controller, a symbol judgment module, a range detection module, a parameter register, a floating point multiplier, a floating point adder, a phase operand arithmetic unit, an address generator, a lookup table, a first gating latch and a second gating latch. The operation of a sigmoid function and a panh function is achieved by configuring a control signal M. The device is simple in achieving structure, the synchronous clock design is adopted, the time sequence checking and the verification are convenient, the area is small, the power consumption is low, the implementation on a chip is convenient, and the practicability of the embedded application is enhanced. When the device is used for calculating the neural network activation function, the processing flow is simple and easy to control, and the calculation efficiency of the neural network activation function is improved. The device can conveniently expand the address generator module and the lookup table module to meet the requirement of function precision conversion according to the requirement of tanh calculation precision. Therefore, the method is an ideal structure for realizing the embedded neural network processor activation function.