東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

A neural network investigation of th...

Smith, Michael A.

Linked to FindBook

Google Book

Amazon

博客來

A neural network investigation of the visuomotor transformation for saccades.

Record Type:	Electronic resources : Monograph/item
Title/Author:	A neural network investigation of the visuomotor transformation for saccades./
Author:	Smith, Michael A.
Description:	206 p.
Notes:	Source: Dissertation Abstracts International, Volume: 64-08, Section: B, page: 4092.
Contained By:	Dissertation Abstracts International64-08B.
Subject:	Psychology, Experimental. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=NQ82826
ISBN:	0612828263

A neural network investigation of the visuomotor transformation for saccades.
Smith, Michael A.

A neural network investigation of the visuomotor transformation for saccades. - 206 p.

Source: Dissertation Abstracts International, Volume: 64-08, Section: B, page: 4092.

Thesis (Ph.D.)--York University (Canada), 2003.

The goal of this study was to train artificial neural networks to generate accurate saccades in Listing's plane, and then determine how the hidden units performed the visuomotor transformation.

ISBN: 0612828263Subjects--Topical Terms:

517106
Psychology, Experimental.

A neural network investigation of the visuomotor transformation for saccades.
LDR:03262nmm 2200325 4500 001 1848153
005 20051102155235.5
008 130614s2003 eng d
020 $a 0612828263
035 $a (UnM)AAINQ82826
035 $a AAINQ82826
040 $a UnM $c UnM
100 1 $a Smith, Michael A. $3 1936171
245 1 2 $a A neural network investigation of the visuomotor transformation for saccades.
300 $a 206 p.
500 $a Source: Dissertation Abstracts International, Volume: 64-08, Section: B, page: 4092.
500 $a Adviser: J. D. Crawford.
502 $a Thesis (Ph.D.)--York University (Canada), 2003.
520 $a The goal of this study was to train artificial neural networks to generate accurate saccades in Listing's plane, and then determine how the hidden units performed the visuomotor transformation.
520 $a CHAPTER 1 networks were designed to illuminate the geometric properties of the visuomotor transformation and therefore used a 3D vector for eye position and motor error, and a 2D vector for retinal error input. The networks in CHAPTER 2 were designed to facilitate a more direct comparison to the physiology of the real saccade generator and used a 6 dimensional encoding of eye position which was designed to mimic the push-pull organization of the extra-ocular muscles. The motor output used the same format as eye position. Visual input consisted of a "retinotopic" input map similar to that of the superficial superior colliculus. In addition, the input units used a Gaussian shaped visual receptive field with which to encode target location.
520 $a The hidden layer of the networks in CHAPTER 1 showed that units divided themselves into 4 parallel modules: a dominant "vector-propagation" class (&sim;50% of units) and 3 classes with specific spatial relations between position, visual, and motor tuning. Surprisingly, the vector-propagation class formed an orthogonal coordinate system aligned with Listing's plane. Selective "lesions" confirmed that this class provided the main drive for saccade magnitude and direction, while a balance between activity in the other classes was required for the correct eye position modulation.
520 $a The hidden units of the networks in CHAPTER 2 developed complex visual receptive fields reminiscent of those found in parietal cortex (Hamed et. al., 2001) while visual sensitivity of these units was modified by eye position using a multiplicative gain field mechanism. In addition, a population code for motor tuning was found to counter-shift with eye position.
520 $a We conclude that error-driven learning is sufficient to produce discrete functional modules and explicit coordinate systems, much like those observed in the real saccade generator and that gain fields work by shifting a motor population code from its visual representation in eye-centered coordinates to a motor vector in head-centered coordinates without ever needing to specify target location in craniotopic coordinates.
590 $a School code: 0267.
650 4 $a Psychology, Experimental. $3 517106
650 4 $a Artificial Intelligence. $3 769149
690 $a 0623
690 $a 0800
710 2 0 $a York University (Canada). $3 1017889
773 0 $t Dissertation Abstracts International $g 64-08B.
790 1 0 $a Crawford, J. D., $e advisor
790 $a 0267
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=NQ82826