Abstract

In the current paper the decoding algorithms for motor-related BCI systems for continuous upper limb trajectory prediction are considered. Two methods for the smooth prediction, namely Sobolev and Polynomial Penalized Multi-Way Partial Least Squares (PLS) regressions, are proposed. The methods are compared to the Multi-Way Partial Least Squares and Kalman Filter approaches. The comparison demonstrated that the proposed methods combined the prediction accuracy of the algorithms of the PLS family and trajectory smoothness of the Kalman Filter. In addition, the prediction delay is significantly lower for the proposed algorithms than for the Kalman Filter approach. The proposed methods could be applied in a wide range of applications beyond neuroscience.

Highlights

  • The Brain-Computer Interface (BCI) is a system for converting the brain’s neural activity into commands for external devices [1]

  • The proposed methods could be applied in a wide range of applications beyond neuroscience

  • The optimal number of factors F (PLS, NPLS, Sobolev Penalized Multi-Way PLS (SNPLS), Penalized Multi-Way PLS (PNPLS)) as well as the smoothing parameter λ (SNPLS, PNPLS) were chosen on the training set by the 10-fold cross-validation procedure [64] for each recording

Read more

Summary

Introduction

The Brain-Computer Interface (BCI) is a system for converting the brain’s neural activity into commands for external devices [1]. Sobolev Penalized Multi-Way PLS (SNPLS) and Polynomial Penalized Multi-Way PLS (PNPLS), are proposed They combine tensor representation of the data with the possibility to control the level of the smoothness of the predicted trajectories. In these methods, the smoothness is not provided by the post-processing or weighting with previous predictions, but is an internal parameter of the identified. A set of state of the art BCI methods, namely Kalman Filter, generic PLS, and Multi-Way PLS (NPLS), are considered and compared with the proposed SNPLS and PNPLS on the set of publicly available ECoG recordings. MADE characterizes the prediction smoothness [29] and, being based on the L1-norm, is robust to the outliers

Methods
Results
Compared methods
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call