NeuroSolutions

survey performed by Juha Ikonen, May 21st 1999

Disclaimer: If any information on this page simply is not true, please tell us about it and we'll correct it ASAP.

Disclaimer: The opinions and observation herein should be considered personal of the person having performed by the survey, at the time of the survey. They do not reflect any official standing of his employer, of the Laboratory of Computer and Information Science or the Neural Networks Research Center.

General

Program name	NeuroSolutions 3.020
Availability	Commercial software, an evaluation version is available at http://www.nd.com/ Prices vary from $195 to $1995 per seat. Company information: NeuroDimension, Inc. 1800 N. Main Street, Suite D4 Gainesville, FL 32609 Phone: 800-ND-IDEAS Outside U.S.: 352-377-5144 Fax: 352-377-9009 email: info@nd.com
Purpose	Neural network simulation environment
Operating system	Windows NT 3.51/4.0 or Windows 95/98
User interface	Graphical user interface, macro language, OLE automation Good in general Good regarding the SOM
Documentation	Extensive: an online help and a manual of over 700 pages.
Other	The program uses an unique symbolism while presenting the structure of a neural network, which may seem cumbersome at first User can add functionality to the program by creating custom DLL:s The program can create ANSI-compatible C++ source code of a neural network which can be integrated into a C++ application.

SOM features

map parameters
Teaching algorithm	[standard/batch/other] [is implementation correct?]
Map size	1- or 2-dimensional map grid [biggest/smallest possible]
Map lattice and shape	both rectangular
Neighborhood function	Function type: line (two nearest nodes for 1D), diamond (four nearest nodes for 2D) or square (eight nearest nodes for 2D)
	Neighborhood size (h): linear, exponential or logarithmic Parameters: minimum and maximum values, update coefficient
	Learning rate (alpha): linear, exponential or logarithmic Parameters: minimum and maximum values, update coefficient
Initialization	Uniformely distributed random values
Distance function	Euclidian, dot product or box car (distance on a grid)
Unknown components	[allowed or not]
Teaching length	In epochs with stopping conditions
efficiency
Speed [Windows NT 4.0, 200 MHz Pentium MMX, 128 MB RAM]	[time for standard run]
Results	[normal/strange] [quantization error, topographic error]
Other	The SOM algorithm uses a conscience bias to determine the winning node, conscience parameters can be set by user Synapse responses can be delayed

[Comments on SOM implementation]

Usability

preprocessing
Input formats	[ascii/XLS/other]
Data handling and selection	[scaling, histograms] [filters, conditions] [flexibility,usability]
postprocessing
Output formats	[ascii/XLS/other]
Map measures	[quantization error,topology error]
Labeling	[no/simple/advanced]
Clustering	[no/simple/advanced/by visualization]
visualization
Inspection of neurons	[no/simple/advanced]
Clusters/map shape	[u-matrix/clusters/projections (Sammon)/...]
Correlations	[component planes/other]
Data projections	[no/single/groups/advanced]
Markers	[labels/...]

http://www.cis.hut.fi/projects/somtoolbox/links/neurosolutions.shtml
somtlbx@mail.cis.hut.fi
Monday, 09-Oct-2000 12:53:09 EEST