1 Citation (Scopus)

Abstract

This paper presents a method for asynchronous decision making using recorded neural data in a binary decision task. This is a demonstration of a technique for developing motor cortical neural prosthetics that do not rely on external cued timing information. The system presented in this paper uses support vector machines and leaky integrate-and-fire elements to predict directional paddle presses. In addition to the traditional metrics of accuracy, asynchronous systems must also optimize the time needed to make a decision. The system presented is able to predict paddle presses with a median accuracy of 88% and all decisions are made before the time of the actual paddle press. An alternative bit rate measure of performance is defined to show that the system proposed here is able to perform the task with the same efficiency as the rats.

Original languageEnglish (US)
Pages (from-to)363-373
Number of pages11
JournalJournal of Neural Engineering
Volume5
Issue number4
DOIs
StatePublished - 2008

Fingerprint

Prosthetics
Support vector machines
Rats
Decision Making
Fires
Demonstrations
Decision making
Support Vector Machine

ASJC Scopus subject areas

  • Biomedical Engineering
  • Cellular and Molecular Neuroscience

Cite this

Asynchronous decision making in a memorized paddle pressing task. / Dankert, James R.; Olson, Byron; Si, Jennie.

In: Journal of Neural Engineering, Vol. 5, No. 4, 2008, p. 363-373.

Research output: Contribution to journalArticle

Dankert, James R. ; Olson, Byron ; Si, Jennie. / Asynchronous decision making in a memorized paddle pressing task. In: Journal of Neural Engineering. 2008 ; Vol. 5, No. 4. pp. 363-373.
@article{1995ca4ae8e44e68b1e090e51cabcd21,
title = "Asynchronous decision making in a memorized paddle pressing task",
abstract = "This paper presents a method for asynchronous decision making using recorded neural data in a binary decision task. This is a demonstration of a technique for developing motor cortical neural prosthetics that do not rely on external cued timing information. The system presented in this paper uses support vector machines and leaky integrate-and-fire elements to predict directional paddle presses. In addition to the traditional metrics of accuracy, asynchronous systems must also optimize the time needed to make a decision. The system presented is able to predict paddle presses with a median accuracy of 88{\%} and all decisions are made before the time of the actual paddle press. An alternative bit rate measure of performance is defined to show that the system proposed here is able to perform the task with the same efficiency as the rats.",
author = "Dankert, {James R.} and Byron Olson and Jennie Si",
year = "2008",
doi = "10.1088/1741-2560/5/4/001",
language = "English (US)",
volume = "5",
pages = "363--373",
journal = "Journal of Neural Engineering",
issn = "1741-2560",
publisher = "IOP Publishing Ltd.",
number = "4",

}

TY - JOUR

T1 - Asynchronous decision making in a memorized paddle pressing task

AU - Dankert, James R.

AU - Olson, Byron

AU - Si, Jennie

PY - 2008

Y1 - 2008

N2 - This paper presents a method for asynchronous decision making using recorded neural data in a binary decision task. This is a demonstration of a technique for developing motor cortical neural prosthetics that do not rely on external cued timing information. The system presented in this paper uses support vector machines and leaky integrate-and-fire elements to predict directional paddle presses. In addition to the traditional metrics of accuracy, asynchronous systems must also optimize the time needed to make a decision. The system presented is able to predict paddle presses with a median accuracy of 88% and all decisions are made before the time of the actual paddle press. An alternative bit rate measure of performance is defined to show that the system proposed here is able to perform the task with the same efficiency as the rats.

AB - This paper presents a method for asynchronous decision making using recorded neural data in a binary decision task. This is a demonstration of a technique for developing motor cortical neural prosthetics that do not rely on external cued timing information. The system presented in this paper uses support vector machines and leaky integrate-and-fire elements to predict directional paddle presses. In addition to the traditional metrics of accuracy, asynchronous systems must also optimize the time needed to make a decision. The system presented is able to predict paddle presses with a median accuracy of 88% and all decisions are made before the time of the actual paddle press. An alternative bit rate measure of performance is defined to show that the system proposed here is able to perform the task with the same efficiency as the rats.

UR - http://www.scopus.com/inward/record.url?scp=59649126516&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=59649126516&partnerID=8YFLogxK

U2 - 10.1088/1741-2560/5/4/001

DO - 10.1088/1741-2560/5/4/001

M3 - Article

C2 - 18784389

AN - SCOPUS:59649126516

VL - 5

SP - 363

EP - 373

JO - Journal of Neural Engineering

JF - Journal of Neural Engineering

SN - 1741-2560

IS - 4

ER -