[Lign251] q 5.7.b

[Roger Levy]

Simone Hartung wrote:
> *Hi Roger,*
> *
> *
> *I do not understand your answer to Leslie's question and I had the same
> question. *
> *
> *
> *
>>  have a question about the homework (question 5.7, part (b)).
>>
>> Hypothesis 1 has only 1 parameter pi0, while hypothesis has 2 parameters
>> piA and piB right? If we only need to change the priors for piA and piB,
>> why should that affect hypothesis 1?
>>
> 
> Actually Hypothesis 2 has 3 parameters -- pi0, piA, and piB.  (pi0
> determines the distribution over outcomes for the first phoneme of the
> sequence.)  So the idea is to leave the prior for pi0 as Beta(1,1) and
> to vary the priors for piA and piB.
> 
> *
> *in part b of 5.7. - did you intend to say H2 instead of H1? *

Hi Simone,

I *think* I see the confusion.  Even though I'm asking for P(H_1 | y),
this value will change when you change the priors for the parameters of
H_2.  Take a look at equations 5.2 and 5.3 in the lecture notes, and
recall that P(y | H_2) will appear as part of P(y).

Does this make sense?

Best

Roger

-- 

Roger Levy                      Email: rlevy at ling.ucsd.edu
Assistant Professor             Phone: 858-534-7219
Department of Linguistics       Fax:   858-534-4789
UC San Diego                    Web:   http://ling.ucsd.edu/~rlevy