Mail Archive Search: <+from:Horace

Results:

References: [ +from:Horace_Tso@pgn.com: 8 ]

Total 8 documents matching your query.

1. Bootstrapping Neural Nets to get CI (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Mon, 09 Aug 2004 10:01:48 -0700; Dear list, Has anyone tried to get some sort of confidence band on a neural net (say, a single hidden layer with a typical sigmoid and a weight decay penalty set by cross-validation) by bootstrapping; /archives/html/s-news/2004-08/msg00052.html (7,148 bytes)
2. Re: Bootstrapping Neural Nets to get CI (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Mon, 09 Aug 2004 10:57:58 -0700; i just want CIs for prediction. I think there is no significance to know the distributions of the connection weights (though it could be interesting by itself). Here is the difficulty i ran into. I d; /archives/html/s-news/2004-08/msg00054.html (9,018 bytes)
3. Making sense of factanal's uniqueness (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Tue, 13 Apr 2004 15:52:32 -0700; Dear list, i'm trying to make sense of the variance of the specific/unique factors from a factanal model. To demonstrate, suppose fdata <- matrix( rnorm(100*5, mean=1, sd=2), nrow=100, ncol=5) factmo; /archives/html/s-news/2004-04/msg00059.html (6,919 bytes)
4. Strange problem (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Thu, 04 Mar 2004 13:24:54 -0800; dear list, i have one fine little script which calls the following, BigM <- function( littleM, n) { bigM <- matrix(0, nrow=n*nrow(littleM), ncol=ncol(littleM)) for ( i in 1:ncol(littleM)) { bigM[,i]; /archives/html/s-news/2004-03/msg00050.html (8,017 bytes)
5. Re: Strange problem (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Thu, 04 Mar 2004 16:28:26 -0800; Yes the matrix is close to 1 Gb (492^3 x 8 byte ~ 953 Mb). Other than the physical memory, the machine has a 4 Gb swap pagefile. I don't know enough about Windows' memory management to say how swapp; /archives/html/s-news/2004-03/msg00054.html (9,720 bytes)
6. Re: probability question (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Mon, 21 Jul 2003 16:02:24 -0700; d; /archives/html/s-news/2003-07/msg00148.html (7,597 bytes)
7. Merging Time Series (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Thu, 19 Jun 2003 18:24:09 -0700; Hi folks, Using 6.1 under Win2K, I want to do a Union of two Time Series. Sounds simple? I've tried every trick out there and nothing works. SPlus support suggests using merge(), but that doesn't wor; /archives/html/s-news/2003-06/msg00139.html (9,950 bytes)
8. Recreating correlation (score: 1): Author: "Horace Tso" <Horace_Tso@pgn.com>; Date: Tue, 24 Jun 2003 13:45:11 -0700; Hi folks, Suppose i have a random vector A of size n. I want to generate a vector B such that correlation between A and B is (exactly) a constant, ie. corr( A, B ) = d, where d is given number. This; /archives/html/s-news/2003-06/msg00178.html (7,586 bytes)

Current List: 1 - 8
Page: [1]