Abstract:The task of blind source separation (BSS) involves separating sources from a mixture without prior knowledge of the sources or the mixing system. Single-channel mixtures and non-linear mixtures are a particularly challenging problem in BSS. In this paper, we propose a novel method for addressing BSS with single-channel non-linear mixtures by leveraging the natural feature subspace specialization ability of multi-encoder autoencoders. During the training phase, our method unmixes the input into the separate encoding spaces of the multi-encoder network and then remixes these representations within the decoder for a reconstruction of the input. Then to perform source inference, we introduce a novel encoding masking technique whereby masking out all but one of the encodings enables the decoder to estimate a source signal. To this end, we also introduce a sparse mixing loss that encourages sparse remixing of source encodings throughout the decoder and a so-called zero reconstruction loss on the decoder for coherent source estimations. To analyze and evaluate our method, we conduct experiments on a toy dataset, designed to demonstrate this property of feature subspace specialization, and with real-world biosignal recordings from a polysomnography sleep study for extracting respiration from electrocardiogram and photoplethysmography signals.

Multi-Task Learning for Blind Source Separation

Blind Source Separation of Single-Channel Mixtures via Multi-Encoder Autoencoders

Exploiting Time–Frequency Sparsity for Dual-Sensor Blind Source Separation

Underdetermined Blind Source Separation Using Sparse Coding

Joint Image Separation and Dictionary Learning

Underdetermined Blind Separation by Combining Sparsity and Independence of Sources

Nonlinear signals separation of adaptive natural gradient learning

Research on Blind Source Separation for Mixed Sub-Gaussian and Super-Gaussian Sources

Blind Source Separation with Compressively Sensed Based on SBL

Underdetermined Blind Source Separation Combining Tensor Decomposition and Nonnegative Matrix Factorization

A Compressed Sensing Approach to Blind Separation of Speech Mixture Based on a Two-Layer Sparsity Model

Blind Source Separation for Convolutive Mixtures and Its Application to Machine Vibrations

A New Method Based on a Multi-Stage Algorithm for Blind Source Separation of Convolutive Mixtures

Compressed sensing based underdetermined blind source separation with unsupervised sparse dictionary self-learning

A New Two-Stage Approach to Underdetermined Blind Source Separation using Sparse Representation

Weierstrass Approach to Blind Source Separation of Multiple Nonlinearly Mixed Signals

Frequency Decomposition Method for Under-determined Blind Source Separation by Complementary Filters

Blind Source Separation Based on Mixed Integer Programming

Underdetermined Blind Source Separation of Speech Mixtures Unifying Dictionary Learning and Sparse Representation.

New Algorithm of Blind Sparse Source Separation

SA-UCBSS: Sparsity-Based Adaptive Underdetermined Convolutive Blind Source Separation