A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces

S. P. Kim, Justin C. Sanchez, Y. N. Rao, D. Erdogmus, J. M. Carmena, M. A. Lebedev, M. A L Nicolelis, J. C. Principe

Research output: Contribution to journalArticle

82 Citations (Scopus)

Abstract

The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) applied to datasets from two experiments in monkeys performing motor tasks (reaching for food and target hitting). Ensembles of 100-200 cortical neurons were simultaneously recorded in these experiments, and even larger neuronal samples are anticipated in the future. Due to the large size of the models (thousands of parameters), the major issue studied was the generalization performance. Every parameter of the models (not only the weights) was selected optimally using signal processing and machine learning techniques. The models were also compared statistically with respect to the Wiener filter as the baseline. Each of the optimization procedures produced improvements over that baseline for either one of the two datasets or both.

Original languageEnglish
Article number009
Pages (from-to)145-161
Number of pages17
JournalJournal of Neural Engineering
Volume3
Issue number2
DOIs
StatePublished - Jun 1 2006
Externally publishedYes

Fingerprint

Brain-Computer Interfaces
Nonlinear Dynamics
Motor Cortex
MIMO systems
Linear Models
Brain
Biomechanical Phenomena
Haplorhini
Hand
Neurons
Weights and Measures
Food
Adaptive filters
Learning systems
Time delay
Signal processing
Kinematics
Animals
Experiments
Datasets

ASJC Scopus subject areas

  • Biotechnology
  • Bioengineering
  • Neuroscience (miscellaneous)

Cite this

Kim, S. P., Sanchez, J. C., Rao, Y. N., Erdogmus, D., Carmena, J. M., Lebedev, M. A., ... Principe, J. C. (2006). A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces. Journal of Neural Engineering, 3(2), 145-161. [009]. https://doi.org/10.1088/1741-2560/3/2/009

A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces. / Kim, S. P.; Sanchez, Justin C.; Rao, Y. N.; Erdogmus, D.; Carmena, J. M.; Lebedev, M. A.; Nicolelis, M. A L; Principe, J. C.

In: Journal of Neural Engineering, Vol. 3, No. 2, 009, 01.06.2006, p. 145-161.

Research output: Contribution to journalArticle

Kim, SP, Sanchez, JC, Rao, YN, Erdogmus, D, Carmena, JM, Lebedev, MA, Nicolelis, MAL & Principe, JC 2006, 'A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces', Journal of Neural Engineering, vol. 3, no. 2, 009, pp. 145-161. https://doi.org/10.1088/1741-2560/3/2/009
Kim SP, Sanchez JC, Rao YN, Erdogmus D, Carmena JM, Lebedev MA et al. A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces. Journal of Neural Engineering. 2006 Jun 1;3(2):145-161. 009. https://doi.org/10.1088/1741-2560/3/2/009
Kim, S. P. ; Sanchez, Justin C. ; Rao, Y. N. ; Erdogmus, D. ; Carmena, J. M. ; Lebedev, M. A. ; Nicolelis, M. A L ; Principe, J. C. / A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces. In: Journal of Neural Engineering. 2006 ; Vol. 3, No. 2. pp. 145-161.
@article{6fbde83df607428a9b83c8406eec14bf,
title = "A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces",
abstract = "The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) applied to datasets from two experiments in monkeys performing motor tasks (reaching for food and target hitting). Ensembles of 100-200 cortical neurons were simultaneously recorded in these experiments, and even larger neuronal samples are anticipated in the future. Due to the large size of the models (thousands of parameters), the major issue studied was the generalization performance. Every parameter of the models (not only the weights) was selected optimally using signal processing and machine learning techniques. The models were also compared statistically with respect to the Wiener filter as the baseline. Each of the optimization procedures produced improvements over that baseline for either one of the two datasets or both.",
author = "Kim, {S. P.} and Sanchez, {Justin C.} and Rao, {Y. N.} and D. Erdogmus and Carmena, {J. M.} and Lebedev, {M. A.} and Nicolelis, {M. A L} and Principe, {J. C.}",
year = "2006",
month = "6",
day = "1",
doi = "10.1088/1741-2560/3/2/009",
language = "English",
volume = "3",
pages = "145--161",
journal = "Journal of Neural Engineering",
issn = "1741-2560",
publisher = "IOP Publishing Ltd.",
number = "2",

}

TY - JOUR

T1 - A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces

AU - Kim, S. P.

AU - Sanchez, Justin C.

AU - Rao, Y. N.

AU - Erdogmus, D.

AU - Carmena, J. M.

AU - Lebedev, M. A.

AU - Nicolelis, M. A L

AU - Principe, J. C.

PY - 2006/6/1

Y1 - 2006/6/1

N2 - The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) applied to datasets from two experiments in monkeys performing motor tasks (reaching for food and target hitting). Ensembles of 100-200 cortical neurons were simultaneously recorded in these experiments, and even larger neuronal samples are anticipated in the future. Due to the large size of the models (thousands of parameters), the major issue studied was the generalization performance. Every parameter of the models (not only the weights) was selected optimally using signal processing and machine learning techniques. The models were also compared statistically with respect to the Wiener filter as the baseline. Each of the optimization procedures produced improvements over that baseline for either one of the two datasets or both.

AB - The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) applied to datasets from two experiments in monkeys performing motor tasks (reaching for food and target hitting). Ensembles of 100-200 cortical neurons were simultaneously recorded in these experiments, and even larger neuronal samples are anticipated in the future. Due to the large size of the models (thousands of parameters), the major issue studied was the generalization performance. Every parameter of the models (not only the weights) was selected optimally using signal processing and machine learning techniques. The models were also compared statistically with respect to the Wiener filter as the baseline. Each of the optimization procedures produced improvements over that baseline for either one of the two datasets or both.

UR - http://www.scopus.com/inward/record.url?scp=33744901439&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33744901439&partnerID=8YFLogxK

U2 - 10.1088/1741-2560/3/2/009

DO - 10.1088/1741-2560/3/2/009

M3 - Article

C2 - 16705271

AN - SCOPUS:33744901439

VL - 3

SP - 145

EP - 161

JO - Journal of Neural Engineering

JF - Journal of Neural Engineering

SN - 1741-2560

IS - 2

M1 - 009

ER -