I have just returned from "Odyssey04: The Speaker and Language Recognition Workshop" (http://www.odyssey04.org) where I presented my first paper that makes use...
... When I tried to do similar things in the context of speech (NOT speaker) recognition (specifically, choosing the number of mixture components to use in a...
... Just a quick comment on the use of the term "assumption" and the concept of the "truth" of this assumption: The best thing for me about probability theory...
... But that is not, in fact, our state of knowledge. We KNOW that frames are correlated, because of the overlap of the feature-extraction windows and the...
... I'm 100% agreed. ... Two comments here: 1. In the speaker detection literature, the score is almost never interpreted as an odds ratio or directly used to...
Hi there, I don't know if this falls within the interests of this group, but here goes. I have a question about a particular set of probability distributions....
104
Lars Johansson
lars.johansson@...
Aug 19, 2004 7:54 pm
Hi! Looking through the draft manuscript of "Probability theory - the logic of science" and comparing with the published book there are several intended...
I would suggest two sources: - bayes.wustl.edu (follow link to Jaynes papers) - The book _E. T. Jaynes: Papers on Probability, Statistics and Statistical...
In addition to the authors pointed to at http://bayes.wustl.edu/, I can highly recommend Tom Minka. He has a nice collection on-line of tutorials at: ...
In different places in PTTLOS Jaynes discusses uninformative priors for theta = probability of success: The Laplace prior is uniform over theta: p_L(theta) =...
... So ignore the bit about Jeffreys' prior. Niko Brummer This e-mail is sent on the Spescom Terms and Conditions that can be accessed by clicking on this link...
p_L(phi) = 1/(2+exp(phi)+exp(-phi)) if |phi| >> 0 then log(p_L(phi)) = -|phi| This e-mail is sent on the Spescom Terms and Conditions that can be accessed by...
Has anybody on this group heard of Phillip Calabrese and conditonal logic. His papers on a non-distributive logic for conditional statements seems like...
Hi all, This may be off topic for this list, but I was wondering if anyone could help me solve this problem: Practical version: "I have a continuous scalar...
You could try a re-paremeterization and work instead with the distribution for y = logit(x) = log(x)-log(1-x) instead. The logit transformation maps [0,1] to...
... 1. Sorry, I missed that detail, BUT ... 2. I think that you could think more productively about this problem if you follow Jaynes's suggestion and change...
Ok, let me give some more detail - I did not originally include this detail because I think without it the question is simpler but still valid: What I really...
1. OK, I understand, your estimand is a model. For scientists models (theories) are important and therefore it is nice to be able to make a hard decision: "The...
Hi Niko, ... Hmm, a debatable point - I don't see a clear-cut distinction between unobserved things which conceivably could be observed and unobserved things...
... I have to disagree with this statement, and I think Jaynes would strongly disagree with it too. This is frequentist thinking, and could also be considered...
Hi Kevin, ... Hmm, I was wondering if someone would accuse me of this - allow me to disagree... ... The question is not whether the data we obtained from...
... Ok, Konrad, this is a constructive conversation, because you have helped me change my mind. I was overly conservative to not want to consider the ...
I want to mention something which has intrigued me for a while. Please see the attached photograph of Support Vector Machine (SVM) guru Vapnik (found at...
P.S. Sorry, the attachment was blocked. The photograph is here: http://yann.lecun.com/ex/images/allyourbayes.jpg ... This e-mail is sent on the Spescom Terms...
... Good question. Perhaps he is just referring to Bayes' rule (as accepted by frequentists) rather than the Bayesian approach? ... Don't know if you're...
... I'm glad you've found it helpful - I on the other hand am losing track of what exactly we are debating here, and I don't see why you are describing my...
... As I recall, you wanted to do model comparison for a nested model. Have you tried the Carlin and Chib approach? You can find an example of doing this...
... I haven't much of really deep philosophical import to add here, but on the "heuristic" side of things, something that's missing from your list is a...
