Compressing deep quaternion neural networks with targeted regularization
This work addresses the need for compact and efficient QVNNs in real-world and embedded applications, representing an incremental improvement in regularization methods for hyper-complex networks.
The paper tackles the problem of overfitting and model size in quaternion neural networks (QVNNs) by designing targeted regularization strategies that minimize connections and neurons during training, resulting in small networks that outperform classical real-valued approaches and are suitable for low-power applications.
In recent years, hyper-complex deep networks (such as complex-valued and quaternion-valued neural networks) have received a renewed interest in the literature. They find applications in multiple fields, ranging from image reconstruction to 3D audio processing. Similar to their real-valued counterparts, quaternion neural networks (QVNNs) require custom regularization strategies to avoid overfitting. In addition, for many real-world applications and embedded implementations, there is the need of designing sufficiently compact networks, with few weights and neurons. However, the problem of regularizing and/or sparsifying QVNNs has not been properly addressed in the literature as of now. In this paper, we show how to address both problems by designing targeted regularization strategies, which are able to minimize the number of connections and neurons of the network during training. To this end, we investigate two extensions of l1 and structured regularization to the quaternion domain. In our experimental evaluation, we show that these tailored strategies significantly outperform classical (real-valued) regularization approaches, resulting in small networks especially suitable for low-power and real-time applications.