Fully complex-valued deep learning model for visual perception
This work addresses the challenge of efficiently leveraging complex-valued data in visual perception tasks, offering a novel method that improves performance and reduces model complexity, though it is incremental in advancing complex-valued deep learning.
The paper tackles the problem of information loss in complex-valued deep learning models by proposing a fully complex-valued convolutional neural network (FC-CNN) with a novel training scheme, achieving 4-10% performance gains over real-valued counterparts and state-of-the-art results on CIFAR-100 with 25% fewer parameters.
Deep learning models operating in the complex domain are used due to their rich representation capacity. However, most of these models are either restricted to the first quadrant of the complex plane or project the complex-valued data into the real domain, causing a loss of information. This paper proposes that operating entirely in the complex domain increases the overall performance of complex-valued models. A novel, fully complex-valued learning scheme is proposed to train a Fully Complex-valued Convolutional Neural Network (FC-CNN) using a newly proposed complex-valued loss function and training strategy. Benchmarked on CIFAR-10, SVHN, and CIFAR-100, FC-CNN has a 4-10% gain compared to its real-valued counterpart, maintaining the model complexity. With fewer parameters, it achieves comparable performance to state-of-the-art complex-valued models on CIFAR-10 and SVHN. For the CIFAR-100 dataset, it achieves state-of-the-art performance with 25% fewer parameters. FC-CNN shows better training efficiency and much faster convergence than all the other models.