public static void Main() {
bool reverse = false;
modshogun.init_shogun_with_defaults();
int order = 3;
int gap = 4;
String[] fm_train_dna = Load.load_dna("../data/fm_train_dna.dat");
StringCharFeatures charfeat = new StringCharFeatures(fm_train_dna, EAlphabet.DNA);
StringWordFeatures feats = new StringWordFeatures(charfeat.get_alphabet());
feats.obtain_from_char(charfeat, order-1, order, gap, reverse);
LinearHMM hmm = new LinearHMM(feats);
hmm.train();
hmm.get_transition_probs();
int num_examples = feats.get_num_vectors();
int num_param = hmm.get_num_model_parameters();
for (int i = 0; i < num_examples; i++)
for(int j = 0; j < num_param; j++) {
hmm.get_log_derivative(j, i);
}
double[] out_likelihood = hmm.get_log_likelihood();
double out_sample = hmm.get_log_likelihood_sample();
}
internal static ArrayList run(IList para)
{
bool reverse = false;
modshogun.init_shogun_with_defaults();
int order = (int)((int?)para[0]);
int gap = (int)((int?)para[1]);
string[] fm_train_dna = Load.load_dna("../data/fm_train_dna.dat");
StringCharFeatures charfeat = new StringCharFeatures(fm_train_dna, DNA);
StringWordFeatures feats = new StringWordFeatures(charfeat.get_alphabet());
feats.obtain_from_char(charfeat, order-1, order, gap, reverse);
LinearHMM hmm = new LinearHMM(feats);
hmm.train();
hmm.get_transition_probs();
int num_examples = feats.get_num_vectors();
int num_param = hmm.get_num_model_parameters();
for (int i = 0; i < num_examples; i++)
{
for(int j = 0; j < num_param; j++)
{
hmm.get_log_derivative(j, i);
}
}
DoubleMatrix out_likelihood = hmm.get_log_likelihood();
double out_sample = hmm.get_log_likelihood_sample();
ArrayList result = new ArrayList();
result.Add(hmm);
result.Add(out_sample);
result.Add(out_likelihood);
modshogun.exit_shogun();
return result;
}
internal static HandleRef getCPtr(LinearHMM obj) {
return (obj == null) ? new HandleRef(null, IntPtr.Zero) : obj.swigCPtr;
}
