Identifiability, Subspace Selection and Noisy ICA
نویسنده
چکیده
We consider identifiability and subspace selection for the noisy ICA model. We discuss a canonical decomposition that allows us to decompose the system into a signal and a noise subspace and show that an unbiased estimate of these can be obtained using a standard ICA algorithm. This can also be used to estimate the relevant subspace dimensions and may often be preferable to PCA dimension reduction. Finally we discuss the identifiability issues for the subsequent ‘square’ noisy ICA model after projection.
منابع مشابه
Identifiability Conditions and Subspace Clustering in Sparse BSS
We give general identifiability conditions on the source matrix in Blind Signal Separation problem. They refine some previously known ones. We develop a subspace clustering algorithm, which is a generalization of the k-plane clustering algorithm, and is suitable for separation of sparse mixtures with bigger sparsity (i.e. when the number of the sensors is bigger at least by 2 than the number of...
متن کاملImproving the Performance of ICA Algorithm for fMRI Simulated Data Analysis Using Temporal and Spatial Filters in the Preprocessing Phase
Introduction: The accuracy of analyzing Functional MRI (fMRI) data is usually decreases in the presence of noise and artifact sources. A common solution in for analyzing fMRI data having high noise is to use suitable preprocessing methods with the aim of data denoising. Some effects of preprocessing methods on the parametric methods such as general linear model (GLM) have previously been evalua...
متن کاملSubspace modeling and selection for noisy speech recognition
This paper presents a new subspace modeling and selection approach for noisy speech recognition. In subspace modeling, we develop factor analysis (FA) for representing noisy speech. FA is a data generation model where the common factors are extracted with factor loading matrix and specific factors. We bridge the connection of FA to signal subspace (SS) approach. Interestingly, FA partitions noi...
متن کاملFeature Selection by Independent Component Analysis for Robust Speaker Verification
A robust approach that unifies independent component analysis (ICA) subspace feature selection in connection with the speaker verification (SV) is proposed. ICA subspace provides statistically independent basis that spans the input space of corrupted speech, then the selected independent components are applied to a vector quantizer (VQ) for SV purpose. The Euclidean distance in the feature spac...
متن کاملIdentifiability and Separability of Linear Ica Models Revisited
We prove theorems that ensure identifiability, separability and uniqueness of linear ICA models. The currently used conditions in ICA community are hence extended to wider class of mixing models and source distributions. Examples illustrating the above concepts are presented as well.
متن کامل