Discovery of Single Independent Latent Variable (2021-10-12T00:00:00.000000Z)

TL;DR

This work considers data given as an invertible mixture of two statistically independent components and assumes that one of the components is observed while the other is hidden, and proposes an autoencoder equipped with a discriminator to recover the hidden component.

Abstract

Latent variable discovery is a central problem in data analysis with a broad range of applications in applied science. In this work, we consider data given as an invertible mixture of two statistically independent components and assume that one of the components is observed while the other is hidden. Our goal is to recover the hidden component. For this purpose, we propose an autoencoder equipped with a discriminator. Unlike the standard nonlinear ICA problem, which was shown to be non-identifiable, in the special case of ICA we consider here, we show that our approach can recover the component of interest up to entropy-preserving transformation. We demonstrate the performance of the proposed approach in several tasks, including image synthesis, voice cloning, and fetal ECG extraction.

Authors

Uri Shaham

2 papers

TL;DR

Abstract

Authors

References67 items

A non-invasive multimodal foetal ECG–Doppler dataset for antenatal cardiology research

ESPnet-SE: End-To-End Speech Enhancement and Separation Toolkit Designed for ASR Integration

Disentanglement by Nonlinear ICA with General Incompressible-flow Networks (GIN)

MelGAN: Generative Adversarial Networks for Conditional Waveform Synthesis

Group Latent Embedding for Vector Quantized Variational Autoencoder in Non-Parallel Voice Conversion

Variational Autoencoders and Nonlinear ICA: A Unifying Framework

The Incomplete Rosetta Stone problem: Identifiability results for Multi-view Nonlinear ICA

TTS Skins: Speaker Conversion via ASR

Unsupervised Singing Voice Conversion

Jasper: An End-to-End Convolutional Neural Acoustic Model

An efficient compression of ECG signals using deep convolutional autoencoders

Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations

Adaptive Linear Neuron for Fetal Electrocardiogram Extraction

Batch Effect Removal via Batch-Free Encoding

Nonlinear ICA Using Auxiliary Variables and Generalized Contrastive Learning

A Universal Music Translation Network

Structured Disentangled Representations

Disentangling by Factorising

Isolating Sources of Disentanglement in VAEs

Learning by coincidence: Siamese networks and common variable learning

The Perception-Distortion Tradeoff

Variational Inference of Disentangled Latent Concepts from Unlabeled Observations

Learning Independent Features with Adversarial Nets for Non-linear ICA

Fader Networks: Manipulating Images by Sliding Attributes

Non-Invasive Fetal Monitoring: A Maternal Surface ECG Electrode Placement-Based Novel Approach for Optimization of Adaptive Filter Control Parameters Using the LMS and RLS Algorithms

Nonlinear ICA of Temporally Dependent Stationary Sources

beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework

SUPERSEDED - CSTR VCTK Corpus: English Multi-speaker Corpus for CSTR Voice Cloning Toolkit

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

Unsupervised Feature Extraction by Time-Contrastive Learning and Nonlinear ICA

A practical guide to non-invasive foetal electrocardiogram extraction and analysis

Nonparametric Canonical Correlation Analysis

Learning the geometry of common latent variables using alternating-diffusion

Domain-Adversarial Training of Neural Networks

Adam: A Method for Stochastic Optimization

Auto-Encoding Variational Bayes

Research of fetal ECG extraction using wavelet analysis and adaptive filtering

Deep Canonical Correlation Analysis

Signal Processing

A Novel Technique for Extraction of FECG using Multi Stage Adaptive Filtering

Handbook of Blind Source Separation: Independent Component Analysis and Applications

Measuring Statistical Dependence with Hilbert-Schmidt Norms

Identifiability, separability, and uniqueness of linear ICA models

Image quality assessment: from error visibility to structural similarity

Laplacian Eigenmaps for Dimensionality Reduction and Data Representation

Nonlinear dimensionality reduction by locally linear embedding.

A global geometric framework for nonlinear dimensionality reduction.

Independent component analysis: algorithms and applications

Nonlinear independent component analysis: Existence and uniqueness results

Nonlinear Component Analysis as a Kernel Eigenvalue Problem

A blind source separation technique using second-order statistics

Adaptive noise cancelling: Principles and applications

Information Theory and Statistical Mechanics

LIII. On lines and planes of closest fit to systems of points in space

A Review of Signal Processing Techniques for Non-Invasive Fetal Electrocardiography

The role of the epistemic communities in the 'constitutionalization' of the Internet Governance: the case of the EU High-Level Expert Group on Artificial Intelligence.

Understanding disentangling in beta-VAE

Unpaired image-to-image translation using cycle-consistent adversarial networks

International Conference on Learning Representations (ICLR)

An extension of slow feature analysis for nonlinear blind source separation

An Echo State Neural Network for Foetal ECG Extraction Optimised by Random Search

Latent Dirichlet Allocation

Diffusion maps

The''echo state''approach to analysing and training recurrent neural networks

Physionet: Components of a New Research Resource for Complex Physiologic Signals". Circu-lation Vol

Physionet: components of a new research resource for complex physiologic signals

Analysis of a complex of statistical variables into principal components.

Field of Study

Venue Information

Name

Type

URL

Alternate Names