RE: Hidden nodes???

• Messages sorted by: [ date ][ thread ][ subject ][ author ]
• Next message: : "cascade correlation in batch mode"
• Previous message: : "SNNSv4.1 660603 patterns loading"

------ =_NextPart_000_01BC9440.A9FED800
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Dear Keith,

I believe you have asked the right question. However, there is still no =
simple answer for=20
it. Some advice (for MLP's) could be found from Lippmann (1987) and Pao =
(1989: p.129). My experience is you could go and look for study =
(reported in journals & papers) using similar input-output patterns and =
tryout their structure first (if they have any good suggestions at all). =
Then, you may want to look for something better with your own =
trial-and-error.

cheers,
Jonathan

In case

1. Y.H. Pao, (1989) Adaptive Pattern Recognition and Neural Networks. =
Addison-Wesley.
2. R.P. Lippmann (1987) IEEE ASSP Magazine, Vol. 4, No.2, April 1987, =
pp, 4-22.
----------
From: 	Shree Ganesh Keith Ragoonaden[SMTP:keithr@erdw.ethz.ch]
Sent: 	1997|~7$e18$e $U$E 11:29
To: 	jonchan@hkusub.hku.hk
Subject: 	Hidden nodes???

Dear all,
I have a question to ask about the hidden layer. Asuming that one has an
input layer with X units and an output layer with Y units. What is the
formula that relates the input and output units with those of the hidden
layer? Moreover, what is the criteria for defining the number of hidden
units if one has X input and Y output units?
I would truly appreciate your cooperation on this issue as there seem to =
be
no apparent solution in the SNNS v4.1 manual. I look forward to here =
from
you soon.
Yours sincerely,

Keith Ragoonaden
Computer Scientist/Software Engineer
=3DB5-Palaeontology group, Geologie Institute
Erdwissenschaften (D-ERDW) Department
Sonneggstrasse 5
ETH Zentrum
CH-8092 Zuerich
Switzerland

Tel :    (++41)-1-6323694
=3D46ax:    (++41)-1-6321080





------ =_NextPart_000_01BC9440.A9FED800
Content-Type: application/ms-tnef
Content-Transfer-Encoding: base64
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------ =_NextPart_000_01BC9440.A9FED800--

• Next message: : "cascade correlation in batch mode"
• Previous message: : "SNNSv4.1 660603 patterns loading"