IterativeN-way partial least squares for a binary self-paced brain–computer interface in freely moving animals

Abstract

In this paper a tensor-based approach is developed for calibration of binary self-paced brain-computer interface (BCI) systems. In order to form the feature tensor, electrocorticograms, recorded during behavioral experiments in freely moving animals (rats), were mapped to the spatial-temporal-frequency space using the continuous wavelet transformation. An N-way partial least squares (NPLS) method is applied for tensor factorization and the prediction of a movement intention depending on neuronal activity. To cope with the huge feature tensor dimension, an iterative NPLS (INPLS) algorithm is proposed. Computational experiments demonstrated the good accuracy and robustness of INPLS. The algorithm does not depend on any prior neurophysiological knowledge and allows fully automatic system calibration and extraction of the BCI-related features. Based on the analysis of time intervals preceding the BCI events, the calibration procedure constructs a predictive model of control. The BCI system was validated by experiments in freely moving animals under conditions close to those in a natural environment.

L1-Regularized Multiway Canonical Correlation Analysis for SSVEP-Based BCI

“…That is, The iteration process does not stop until each of , , and satisfies (11). To comprehensively assess efficiencies of the proposed methods, Table II investigates the computational time of the MCCA and the L1-MCCA for reference signal optimization, TABLE II COMPUTATIONAL TIME OF THE MCCA AND THE L1-MCCA FOR REFERENCE SIGNAL OPTIMIZATION, AND THAT OF THE CCA FOR SSVEP RECOGNITION, EVALUATED FROM THE LEAVE-ONE- Collaboratively multiway optimization has been suggested to be more promising than the one-way optimization for EEG data analysis [39]- [43] and also for electrocorticogram (ECoG) data analysis [44], [45]. Recently, regularized tensor discriminant analysis and penalized -way partial least squares have began to emerge and shown great potentials for the BCI application [46], [47].…”

Recursive N-Way Partial Least Squares for Brain-Computer Interface

“…The tests were performed for the binary output variable which corresponds to the binary BCI experiment described below (see details in [8]). Despite its simplicity, this model makes it possible to explore the fundamental properties of the method.…”

“…The model allows the control of an external effector at the stage of the online execution. Multi-way analysis was recently reported [1]–[8] to be an efficient way to calibrate BCI systems by providing simultaneous signal processing in several domains (temporal, frequency and spatial). Multi-way analysis represents a natural approach for modalities fusion [9].…”

“…The algorithm is addressed to a data set of huge dimensions and conducts a sequential blockwise tensor-data processing. Unlike the multi-pass blockwise Iterative NPLS [8], which repeatedly runs through the entire data set, the new RNPLS algorithm performs a consecutive calculation and can be applied online. Moreover, in the case of non-stationary tensor-valued processes, RNPLS allows adaptive learning by introducing the forgetting factor.…”

A New Generation of Brain-Computer Interfaces Driven by Discovery of Latent EEG-fMRI Linkages Using Tensor Decomposition

“…Therefore, we believe that there is initial experimental basis to believe that fMRI-inspired EEG BCIs may be viable. BCI using tensor decomposition techniques has been successfully performed with EEG before (Eliseyev et al, 2011; Eliseyev and Aksenova, 2013), but not with fMRI or with a novel framework like the one we are proposing. The open loop BCI is the focus of this paper.…”