Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network

Justin C. Sanchez, J. C. Principe, J. M. Carmena, Mikhail A. Lebedev, M. A L Nicolelis

Research output: Chapter in Book/Report/Conference proceedingConference contribution

11 Citations (Scopus)

Abstract

Implementation of brain-machine interface neural-to-motor mapping algorithms in low-power, portable digital signal processors (DSPs) requires efficient use of model resources especially when predicting signals that show interdependencies. We show here that a single recurrent neural network can simultaneously predict hand position and velocity from the same ensemble of cells using a minimalist topology. Analysis of the trained topology showed that the model learns to concurrently represent multiple kinematic parameters in a single state variable. We further assess the expressive power of the state variables for both large and small topologies.

Original languageEnglish (US)
Title of host publicationAnnual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings
Pages5321-5324
Number of pages4
Volume26 VII
StatePublished - 2004
Externally publishedYes
EventConference Proceedings - 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2004 - San Francisco, CA, United States
Duration: Sep 1 2004Sep 5 2004

Other

OtherConference Proceedings - 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2004
CountryUnited States
CitySan Francisco, CA
Period9/1/049/5/04

Fingerprint

Recurrent neural networks
Brain
Kinematics
Topology
Digital signal processors

Keywords

  • Brain-Machine Interface
  • Neuroprosthetic
  • Recurrent neural network
  • RMLP
  • State-space analysis

ASJC Scopus subject areas

  • Bioengineering

Cite this

Sanchez, J. C., Principe, J. C., Carmena, J. M., Lebedev, M. A., & Nicolelis, M. A. L. (2004). Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network. In Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings (Vol. 26 VII, pp. 5321-5324)

Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network. / Sanchez, Justin C.; Principe, J. C.; Carmena, J. M.; Lebedev, Mikhail A.; Nicolelis, M. A L.

Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. Vol. 26 VII 2004. p. 5321-5324.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sanchez, JC, Principe, JC, Carmena, JM, Lebedev, MA & Nicolelis, MAL 2004, Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network. in Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. vol. 26 VII, pp. 5321-5324, Conference Proceedings - 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2004, San Francisco, CA, United States, 9/1/04.
Sanchez JC, Principe JC, Carmena JM, Lebedev MA, Nicolelis MAL. Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network. In Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. Vol. 26 VII. 2004. p. 5321-5324
Sanchez, Justin C. ; Principe, J. C. ; Carmena, J. M. ; Lebedev, Mikhail A. ; Nicolelis, M. A L. / Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network. Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings. Vol. 26 VII 2004. pp. 5321-5324
@inproceedings{5e3e0a69b22545c3a6fd04b61dcef2cd,
title = "Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network",
abstract = "Implementation of brain-machine interface neural-to-motor mapping algorithms in low-power, portable digital signal processors (DSPs) requires efficient use of model resources especially when predicting signals that show interdependencies. We show here that a single recurrent neural network can simultaneously predict hand position and velocity from the same ensemble of cells using a minimalist topology. Analysis of the trained topology showed that the model learns to concurrently represent multiple kinematic parameters in a single state variable. We further assess the expressive power of the state variables for both large and small topologies.",
keywords = "Brain-Machine Interface, Neuroprosthetic, Recurrent neural network, RMLP, State-space analysis",
author = "Sanchez, {Justin C.} and Principe, {J. C.} and Carmena, {J. M.} and Lebedev, {Mikhail A.} and Nicolelis, {M. A L}",
year = "2004",
language = "English (US)",
volume = "26 VII",
pages = "5321--5324",
booktitle = "Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings",

}

TY - GEN

T1 - Simultaneous prediction of four kinematic variables for a brain-machine interface using a single recurrent neural network

AU - Sanchez, Justin C.

AU - Principe, J. C.

AU - Carmena, J. M.

AU - Lebedev, Mikhail A.

AU - Nicolelis, M. A L

PY - 2004

Y1 - 2004

N2 - Implementation of brain-machine interface neural-to-motor mapping algorithms in low-power, portable digital signal processors (DSPs) requires efficient use of model resources especially when predicting signals that show interdependencies. We show here that a single recurrent neural network can simultaneously predict hand position and velocity from the same ensemble of cells using a minimalist topology. Analysis of the trained topology showed that the model learns to concurrently represent multiple kinematic parameters in a single state variable. We further assess the expressive power of the state variables for both large and small topologies.

AB - Implementation of brain-machine interface neural-to-motor mapping algorithms in low-power, portable digital signal processors (DSPs) requires efficient use of model resources especially when predicting signals that show interdependencies. We show here that a single recurrent neural network can simultaneously predict hand position and velocity from the same ensemble of cells using a minimalist topology. Analysis of the trained topology showed that the model learns to concurrently represent multiple kinematic parameters in a single state variable. We further assess the expressive power of the state variables for both large and small topologies.

KW - Brain-Machine Interface

KW - Neuroprosthetic

KW - Recurrent neural network

KW - RMLP

KW - State-space analysis

UR - http://www.scopus.com/inward/record.url?scp=11144347817&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=11144347817&partnerID=8YFLogxK

M3 - Conference contribution

VL - 26 VII

SP - 5321

EP - 5324

BT - Annual International Conference of the IEEE Engineering in Medicine and Biology - Proceedings

ER -