Gate-Variants of Gated Recurrent Unit (GRU) Neural Networks

TL;DR

This paper explores three simplified variants of the GRU neural network, reducing parameters in the gates, and demonstrates they perform comparably to the original while being more computationally efficient.

Contribution

The paper introduces and evaluates three parameter-reduced GRU variants, showing they maintain performance with lower computational costs.

Findings

Variants perform as well as original GRU on MNIST and IMDB

Reduced parameters lead to lower computational expense

Maintains accuracy while improving efficiency

Abstract

The paper evaluates three variants of the Gated Recurrent Unit (GRU) in recurrent neural networks (RNN) by reducing parameters in the update and reset gates. We evaluate the three variant GRU models on MNIST and IMDB datasets and show that these GRU-RNN variant models perform as well as the original GRU RNN model while reducing the computational expense.