Deep Complex Networks

Abstract

At present, the vast majority of building blocks, techniques, and architectures for deep learning are based on real-valued operations and representations. However, recent work on recurrent neural networks and older fundamental theoretical analysis suggests that complex numbers could have a richer representational capacity and could also facilitate noise-robust memory retrieval mechanisms. Despite their attractive properties and potential for opening up entirely new neural architectures, complex-valued deep neural networks have been marginalized due to the absence of the building blocks required to design such models. In this work, we provide the key atomic components for complex-valued deep neural networks and apply them to convolutional feed-forward networks and convolutional LSTMs. More precisely, we rely on complex convolutions and present algorithms for complex batch-normalization, complex weight initialization strategies for complex-valued neural nets and we use them in experiments with end-to-end training schemes. We demonstrate that such complex-valued models are competitive with their real-valued counterparts. We test deep complex models on several computer vision tasks, on music transcription using the MusicNet dataset and on Speech Spectrum Prediction using the TIMIT dataset. We achieve state-of-the-art performance on these audio-related tasks.

References in corpus (3)

• On the Properties of Neural Machine Translation: Encoder-Decoder Approaches
• Harmonic Networks: Deep Translation and Rotation Equivariance
• Complex and Holographic Embeddings of Knowledge Graphs: A Comparison

Cited by in corpus (51)

• Deep Learning for Audio Signal Processing
• DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancement
• Complex-valued neural networks for machine learning on non-stationary physical data
• Quantum-inspired Complex Convolutional Neural Networks
• Highly Scalable Image Reconstruction using Deep Neural Networks with Bandpass Filtering
• Stochastic Restoration of Heavily Compressed Musical Audio using Generative Adversarial Networks
• Spatially Regularized Parametric Map Reconstruction for Fast Magnetic Resonance Fingerprinting
• Complex and Holographic Embeddings of Knowledge Graphs: A Comparison
• Real-time Monaural Speech Enhancement With Short-time Discrete Cosine Transform
• Deep Complex Networks for Protocol-Agnostic Radio Frequency Device Fingerprinting in the Wild
• DeepcomplexMRI: Exploiting deep residual network for fast parallel MR imaging with complex convolution
• Lightweight and Efficient Neural Natural Language Processing with Quaternion Networks
• 3D-Rotation-Equivariant Quaternion Neural Networks
• Analysis of Deep Complex-Valued Convolutional Neural Networks for MRI Reconstruction
• Quaternion-Valued Convolutional Neural Network Applied for Acute Lymphoblastic Leukemia Diagnosis
• DeepSense: Enabling Carrier Sense in Low-Power Wide Area Networks Using Deep Learning
• DIMENSION: Dynamic MR Imaging with Both K-space and Spatial Prior Knowledge Obtained via Multi-Supervised Network Training
• Deep Quaternion Features for Privacy Protection
• Signal Transformer: Complex-valued Attention and Meta-Learning for Signal Recognition
• Compressed Sensing: From Research to Clinical Practice with Data-Driven Learning
• MITNet: GAN Enhanced Magnetic Induction Tomography Based on Complex CNN
• Complex-valued Federated Learning with Differential Privacy and MRI Applications
• Traffic flow prediction using Deep Sedenion Networks
• Compressing complex convolutional neural network based on an improved deep compression algorithm
• A Two-stage Complex Network using Cycle-consistent Generative Adversarial Networks for Speech Enhancement
• SurReal: Complex-Valued Learning as Principled Transformations on a Scaling and Rotation Manifold
• Binary Complex Neural Network Acceleration on FPGA
• An End-to-end Architecture of Online Multi-channel Speech Separation
• Complex Unitary Recurrent Neural Networks using Scaled Cayley Transform
• Complex Transformer: A Framework for Modeling Complex-Valued Sequence
• BCNN: Binary Complex Neural Network
• Quaternion Collaborative Filtering for Recommendation
• Unitary Learning for Deep Diffractive Neural Network
• Human Listening and Live Captioning: Multi-Task Training for Speech Enhancement
• Machine Learning aided k-t SENSE for fast reconstruction of highly accelerated PCMR data
• End-to-end speech enhancement based on discrete cosine transform
• Pixel-Wise PolSAR Image Classification via a Novel Complex-Valued Deep Fully Convolutional Network
• Quaternion Convolutional Neural Networks
• Phase Aware Speech Enhancement using Realisation of Complex-valued LSTM
• Learning Site-Specific Probing Beams for Fast mmWave Beam Alignment
• DONet: Dual-Octave Network for Fast MR Image Reconstruction
• Deep Learning Framework for Hybrid Analog-Digital Signal Processing in mmWave Massive-MIMO Systems
• Deep learning estimation of complex reverberant wave fields by a programmable metasurface
• Complex-BP-Neural-Network-based Hybrid Precoding for Millimeter Wave Multiuser Massive MIMO Systems
• MD-Recon-Net: A Parallel Dual-Domain Convolutional Neural Network for Compressed Sensing MRI
• ComplexFace: a Multi-Representation Approach for Image Classification with Small Dataset
• RingCNN: Exploiting Algebraically-Sparse Ring Tensors for Energy-Efficient CNN-Based Computational Imaging
• Utilizing Complex-valued Network for Learning to Compare Image Patches
• Widely Linear Complex-valued Autoencoder: Dealing with Noncircularity in Generative-Discriminative Models
• Backpropagation with N-D Vector-Valued Neurons Using Arbitrary Bilinear Products
• Widely Linear Kernels for Complex-Valued Kernel Activation Functions