Add to Favourites
To login click here
This article discusses the use of Quaternion Neural Networks (QNNs) as an extension of the real Convolutional Neural Network (CNN) model. QNNs are a more recent form of neural network that uses quaternion-valued inputs, activation, and parameters. It is shown that QNNs can outperform their real-valued equivalents in tasks such as image processing and speech recognition. Additionally, QNNs benefit from parameter sharing as a result of the interactions of the Hamilton product, resulting in models that require fewer parameters and less storage space.
