Nonlinear Independent Component Analysis for EEG-Based Brain-Computer Interface Systems

Oveisi, Farid; Oveisi, Shahrzad; Efranian, Abbas; Patras, Ioannis

doi:10.5772/39092

Cited by 4 publications

(2 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Identifiability research of linear mixtures is relatively well established. However, unknown nonlinear distortions happen ubiquitously in practice; see examples in hyperspectral imaging, audio processing, wireless communications, and brain imaging [14][15][16]. Naturally, establishing latent component identifiability in the presence of unknown nonlinear transformations is much more challenging relative to classic linear UML cases.…”

Section: Introductionmentioning

confidence: 99%

Identifiability-Guaranteed Simplex-Structured Post-Nonlinear Mixture Learning via Autoencoder

Lyu,

2021

Preprint

View full text Add to dashboard Cite

This work focuses on the problem of unraveling nonlinearly mixed latent components in an unsupervised manner. The latent components are assumed to reside in the probability simplex, and are transformed by an unknown post-nonlinear mixing system. This problem finds various applications in signal and data analytics, e.g., nonlinear hyperspectral unmixing, image embedding, and nonlinear clustering. Linear mixture learning problems are already ill-posed, as identifiability of the target latent components is hard to establish in general. With unknown nonlinearity involved, the problem is even more challenging. Prior work offered a function equation-based formulation for provable latent component identification. However, the identifiability conditions are somewhat stringent and unrealistic. In addition, the identifiability analysis is based on the infinite sample (i.e., population) case, while the understanding for practical finite sample cases has been elusive. Moreover, the algorithm in the prior work trades model expressiveness with computational convenience, which often hinders the learning performance. Our contribution is threefold. First, new identifiability conditions are derived under largely relaxed assumptions. Second, comprehensive sample complexity results are presented-which are the first of the kind. Third, a constrained autoencoder-based algorithmic framework is proposed for implementation, which effectively circumvents the challenges in the existing algorithm. Synthetic and real experiments corroborate our theoretical analyses.

show abstract

Section: Introductionmentioning

confidence: 99%

Identifiability-Guaranteed Simplex-Structured Post-Nonlinear Mixture Learning via Autoencoder

Lyu,

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Dongwei et al (2013) used the ICA analysis and performed a Granger Casuality Analysis (GCA) to detect the interactive dependencies between independent components of ICA. Oveisi et al (2007) studied the nonlinear Independent Component Analysis for EEG brain computer interface system for Blind Source Separation on healthy volunteer on 10-20 electrode system. Omega Complexity during meditation was compared during pre and post meditation by Faber et al (2011).…”

Section: Introductionmentioning

confidence: 99%

Focused Attention Analysis of Meditating and Non-meditating Brains in Time and Frequency Domains Using EEG Data

Selvaraj¹,

Sivaprakasam²

2014

RJASET

View full text Add to dashboard Cite

The activity and the ability of brain to maintain the state of calmness in individuals practicing meditation has been a subject of research from long time. The aim of the study here is to prove that the meditation aids in retaining the state of calmness of brain. A MATLAB based multifaceted framework is developed for analyzing the dataset of brain EEG of people practicing meditation. The proposed method performs the processing of 32 electrode EEG data and denoises the signal in time series. The plotting of data followed by PSD analysis and FFT transform of the signal to analyze the data in frequency domain for examining each frequency band. The comparison is done using the L2 norm. The ICWT is later found to analyze the data and calculate for Modulus and angle of the EEG signal. The statistical analysis in time and frequency domain is use to study the effect of meditation on focused attention and retaining of same in meditating and non-meditating brains.

show abstract

Nonlinear Multiview Analysis: Identifiability and Neural Network-Assisted Implementation

Lyu

2020

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

Nonlinear Independent Component Analysis for EEG-Based Brain-Computer Interface Systems

Cited by 4 publications

References 25 publications

Identifiability-Guaranteed Simplex-Structured Post-Nonlinear Mixture Learning via Autoencoder

Identifiability-Guaranteed Simplex-Structured Post-Nonlinear Mixture Learning via Autoencoder

Focused Attention Analysis of Meditating and Non-meditating Brains in Time and Frequency Domains Using EEG Data

Nonlinear Multiview Analysis: Identifiability and Neural Network-Assisted Implementation

Contact Info

Product

Resources

About