I am trying to extend my Hidden-Markov-Model from 4 internal states and 4 observations to 8 states and 8 observations (currently they have pretty much the same meaning, but I might reduce number of internal states).
But now I am getting the error:
The number of 8 emission paramters for state 0 is invalid:
State order can not be determined.
I can find the error in the source code, but I do not understand where there should a problem. From a theortical point of view I have always been able to add observations by just adding a new observation and a new emission probability for each state. I never heard of something like an order in hidden markov models (except for the order of a markov chain, but we always assume 1 there and it doesn't have anything to do with observations).
Does anyone know what's the problem here and how I can solve it?
Make sure you have a matching number of output symbols (in examples called sigma) to the number of elements in your emission probabiliy matrix (in the examples called B).
I forgot to increment the number of elements in sigma.
Related
I work in theoretical physics, and I have come upon a problem that requires the minimization of a particular Hamiltonian operator for a system of 8 particles, with one non-linear constraint. Due to the complexity of the system, I cannot define the entire Hamiltonian "in one go", nor the constraint. By this I mean that the quantity I am searching for is defined recurrently, depending on complex summations over quantities calculated for systems of 7 particles, which in turn depend on quantities calculated for systems of 6, and so on, until it reaches a one or two-particle system, for which said quantities are given as initial values, dependent on the elements of a column vector (the argument/minization parameters). The constraint itself is also of this form, requiring the "overlap" between the states of 8 particles to be exactly 1. (I.E. the state be normalized) I have been thinking of a way to use fmincon for this, but I've come up short, since my function has an implicit dependence on the parameters, and I can't write the whole thing explicitly. For a better understanding, here is some of the code:
for m=3:npairs+1
for n=3:npairs+1
for i=1:nsps
for j=1:nsps
overlap(m,n)=overlap(m,n)+x(i)*x(j)*(delta(i,j)*(overlap(m-1,n-1)-N(m-1,n-1,i))+p0p(m-1,n-1,j,i));
p(m,n,i)=(n-1)*x(i)*overlap(m,n-1)-(n-2)*(n-1)*x(i)*x(i)*((m-1)*x(i)*overlap(m-1,n-1)-(m-2)*(m-1)*x(i)*x(i)*p(m-1,n-1,i));
N(m,n,i)=2*(n-1)*x(i)*p(n-1,m,i);
p0p(m,n,i,j)=(m-1)*(n-1)*x(i)*x(j)*overlap(m-1,n-1)-(m-1)*(n-1)*(m-2)*x(i)*x(i)*x(j)*p(m-2,n-1,i)-(m-1)*(n-1)*(n-2)*x(i)*x(j)*x(j)*p0(m-1,n-2,j)-(m-1)*(n-1)*(m-2)*(n-2)*x(i)*x(i)*x(j)*x(j)*(delta(i,j)*(overlap(m-2,n-2)-N(m-2,n-2,i))+p0p(m-2,n-2,j,i));
endfor
endfor
endfor
endfor
function [E]=H(x)
E=summation over all i and j of N and p0p for m=n=8 %not actual code
endfunction
overlap(9,9)=1 %constraint
It's hard to give a specific answer, but I would advise the following to get you started.
First, note that, the inner two steps of the nest loop can be vectorised, since i and j always appear as indices (whereas m and n make backreferences, so they cannot be vectorised). So your 4-level loop can be reduced to a 2-level loop containing 4 functions operating over i-by-j matrices.
Second, note that the whole construct can be expressed as a recursive function. If you have suitable base cases for m = 0, n = 0, you can iteratively obtain all i,j matrices for all cases up to m=9,n=9. In particular, you can try to 'memoize' the early steps, and plug them into higher steps, rather than rely on actual recursion.
Assuming you need to sum with the first two indeces fixed to 8 (if I understood correctly), you can easily do with Anonymous Functions
https://octave.org/doc/v6.1.0/Anonymous-Functions.html#Anonymous-Functions
# creating same data
A=ones(8,8,4,4);
B=2*ones(8,8,4,4);
# defining 2 versions of sums
f = #(A,B) [sum(sum(A(8,8,:,:))), sum(sum(B(8,8,:,:)))];
g = #(A,B) sum(sum(A(8,8,:,:)))+ sum(sum(B(8,8,:,:)));
E1=f(A,B)
E2=g(A,B)
the output will be:
octave:21> E1=f(A,B)
E1 =
16 32
octave:22> E2=g(A,B)
E2 = 48
I am trying to get to get familiar with working with a 74LS148 Priority Encoder IC. I am providing a 5 Volt constant current voltage source to the IC and have a 0 Volt ground Voltage. I tried connecting every input to the 5 Volt source to set it on a HIGH logic state, which should give, according to the truth table, HIGH in the output. Then tried setting some inputs to LOW, but that had no effect in the output which remained in a HIGH state.
I then tried using pull up resistors, the use and circuit configuration of which is not entirely clear to me, which I think is the problem. I connected the resistors as shown in the picture below, which should give a HIGH state in the output. I then tried connecting some inputs, along with their resistors, to the Ground. The output still remained on a HIGH state, around 4.3 Volts.
I repeated the entire process with another 74LS148 IC to make sure the first one was working.
I'd could really use a little help. Thank you all!
74LS148 has the following truth table:
EI (pin-5) must low (connect to ground)
The enable E1 must be ground. you also your diagram indicates inputs 0 through 7 are pulled high. This will always produce your outputs to be HIGH
I think you should include a bleeder resistor at output to draw some constant current. This will help regulator function as desired.
I am a starter in text mining topic. When I run LDA() over a huge dataset with 996165 observations, it displays the following error:
Error in LDA(dtm, k, method = "Gibbs", control = list(nstart = nstart, :
Each row of the input matrix needs to contain at least one non-zero entry.
I am pretty sure that there is no missing values in my corpus and also. The table of "DocumentTermMatrix" and "simple_triplet_matrix" is:
table(is.na(dtm[[1]]))
#FALSE
#57100956
table(is.na(dtm[[2]]))
#FALSE
#57100956
A little confused how "57100956" comes. But as my dataset is pretty large, I don't know how to check why does this error occurs. My LDA command is:
ldaOut<-LDA(dtm,k, method="Gibbs", control=list(nstart=nstart, seed = seed, best=best, burnin = burnin, iter = iter, thin=thin))
Can anyone provide some insights? Thanks.
In my opinion the problem is not the presence of missing values, but the presence of all 0 rows.
To check it:
raw.sum=apply(table,1,FUN=sum) #sum by raw each raw of the table
Then you can delete all raws which are all 0 doing:
table=table[raw.sum!=0,]
Now table should has all "non 0" raws.
I had the same problem. The design matrix, dtm, in your case, had rows with all zeroes because dome documents did not contain certain words (i.e. their frequency was zero). I suppose this somehow causes a singular matrix problem somewhere along the line. I fixed this by adding a common word to each of the documents so that every row would have at least one non-zero entry. At the very least, the LDA ran successfully and classified each of the documents. Hope this helps!
I am running a wavelet transform (cmor) to estimate damping and frequencies that exists in a signal.cmor has 2 parameters that I can change them to get more accurate results. center frequency(Fc) and bandwidth frequency(Fb). If I construct a signal with few freqs and damping then I can measure the error of my estimation(fig 2). but in actual case I have a signal and I don't know its freqs and dampings so I can't measure the error.so a friend in here suggested me to reconstruct the signal and find error by measuring the difference between the original and reconstructed signal e(t)=|x(t)−x^(t)|.
so my question is:
Does anyone know a better function to find the error between reconstructed and original signal,rather than e(t)=|x(t)−x^(t)|.
can I use GA to search for Fb and Fc? or do you know a better search method?
Hope this picture shows what I mean, the actual case is last one. others are for explanations
Thanks in advance
You say you don't know the error until after running the wavelet transform, but that's fine. You just run a wavelet transform for every individual the GA produces. Those individuals with lower errors are considered fitter and survive with greater probability. This may be very slow, but conceptually at least, that's the idea.
Let's define a Chromosome datatype containing an encoded pair of values, one for the frequency and another for the damping parameter. Don't worry too much about how their encoded for now, just assume it's an array of two doubles if you like. All that's important is that you have a way to get the values out of the chromosome. For now, I'll just refer to them by name, but you could represent them in binary, as an array of doubles, etc. The other member of the Chromosome type is a double storing its fitness.
We can obviously generate random frequency and damping values, so let's create say 100 random Chromosomes. We don't know how to set their fitness yet, but that's fine. Just set it to zero at first. To set the real fitness value, we're going to have to run the wavelet transform once for each of our 100 parameter settings.
for Chromosome chr in population
chr.fitness = run_wavelet_transform(chr.frequency, chr.damping)
end
Now we have 100 possible wavelet transforms, each with a computed error, stored in our set called population. What's left is to select fitter members of the population, breed them, and allow the fitter members of the population and offspring to survive into the next generation.
while not done
offspring = new_population()
while count(offspring) < N
parent1, parent2 = select_parents(population)
child1, child2 = do_crossover(parent1, parent2)
mutate(child1)
mutate(child2)
child1.fitness = run_wavelet_transform(child1.frequency, child1.damping)
child2.fitness = run_wavelet_transform(child2.frequency, child2.damping)
offspring.add(child1)
offspring.add(child2)
end while
population = merge(population, offspring)
end while
There are a bunch of different ways to do the individual steps like select_parents, do_crossover, mutate, and merge here, but the basic structure of the GA stays pretty much the same. You just have to run a brand new wavelet decomposition for every new offspring.
The only effect AudioUnit on iOS is the "iTunes EQ", which only lets you use EQ pre-sets. I would like to use a customized eq in my audio graph
I came across this question on the subject and saw an answer suggesting using this DSP code in the render callback. This looks promising and people seem to be using this effectively on various platforms. However, my implementation has a ton of noise even with a flat eq.
Here's my 20 line integration into the "MixerHostAudio" class of Apple's "MixerHost" example application (all in one commit):
https://github.com/tassock/mixerhost/commit/4b8b87028bfffe352ed67609f747858059a3e89b
Any ideas on how I could get this working? Any other strategies for integrating an EQ?
Edit: Here's an example of the distortion I'm experiencing (with the eq flat):
http://www.youtube.com/watch?v=W_6JaNUvUjA
In the code in EQ3Band.c, the filter coefficients are used without being initialized. The init_3band_state method initialize just the gains and frequencies, but the coefficients themselves - es->f1p0 etc. are not initialized, and therefore contain some garbage values. That might be the reason for the bad output.
This code seems wrong in more then one way.
A digital filter is normally represented by the filter coefficients, which are constant, the filter inner state history (since in most cases the output depends on history) and the filter topology, which is the arithmetic used to calculate the output given the input and the filter (coeffs + state history). In most cases, and of course when filtering audio data, you expect to get 0's at the output if you feed 0's to the input.
The problems in the code you linked to:
The filter coefficients are changed in each call to the processing method:
es->f1p0 += (es->lf * (sample - es->f1p0)) + vsa;
The input sample is usually multiplied by the filter coefficients, not added to them. It doesn't make any physical sense - the sample and the filter coeffs don't even have the same physical units.
If you feed in 0's, you do not get 0's at the output, just some values which do not make any sense.
I suggest you look for another code - the other option is debugging it, and it would be harder.
In addition, you'd benefit from reading about digital filters:
http://en.wikipedia.org/wiki/Digital_filter
https://ccrma.stanford.edu/~jos/filters/