prox's profile - activity

But instead of the pure class value I would like to have the probability of the sample(s) being part of the first class, as long as it is a 2-Class problem.

In recent version we had Random Forests' predict_prob method to do so which I am missing in latest version.

Hi,

I'm looking for the predict probability functionality for binary classification task present in 2.x releases. As far as I see it is gone, is there a tendency towards getting it back?

Best regards, chris

Looks like =)

Hi,

I was looking for updating to 3.0 and checked the new TrainData class and how it works with the SVM TrainAuto. As far as I can see during cross validation, the temp_test_samples matrix is never filled with data or am I getting anything wrong?

Mat temp_test_samples(test_sample_count, var_count, CV_32F);

predict(temp_test_samples, temp_test_responses, 0);

    int test_sample_count = (sample_count + k_fold/2)/k_fold;
    int train_sample_count = sample_count - test_sample_count;

    SvmParams best_params = params;
    double min_error = FLT_MAX;

    int rtype = responses.type();

    Mat temp_train_samples(train_sample_count, var_count, CV_32F);
    Mat temp_test_samples(test_sample_count, var_count, CV_32F);
    Mat temp_train_responses(train_sample_count, 1, rtype);
    Mat temp_test_responses;

    // If grid.minVal == grid.maxVal, this will allow one and only one pass through the loop with params.var = grid.minVal.
    #define FOR_IN_GRID(var, grid) \
        for( params.var = grid.minVal; params.var == grid.minVal || params.var < grid.maxVal; params.var = (grid.minVal == grid.maxVal) ? grid.maxVal + 1 : params.var * grid.logStep )

    FOR_IN_GRID(C, C_grid)
    FOR_IN_GRID(gamma, gamma_grid)
    FOR_IN_GRID(p, p_grid)
    FOR_IN_GRID(nu, nu_grid)
    FOR_IN_GRID(coef0, coef_grid)
    FOR_IN_GRID(degree, degree_grid)
    {
        // make sure we updated the kernel and other parameters
        setParams(params);

        double error = 0;
        for( k = 0; k < k_fold; k++ )
        {
            int start = (k*sample_count + k_fold/2)/k_fold;
            for( i = 0; i < train_sample_count; i++ )
            {
                j = sidx[(i+start)%sample_count];
                memcpy(temp_train_samples.ptr(i), samples.ptr(j), sample_size);
                if( is_classification )
                    temp_train_responses.at<int>(i) = responses.at<int>(j);
                else if( !responses.empty() )
                    temp_train_responses.at<float>(i) = responses.at<float>(j);
            }

            // Train SVM on <train_size> samples
            if( !do_train( temp_train_samples, temp_train_responses ))
                continue;

            for( i = 0; i < train_sample_count; i++ )
            {
                j = sidx[(i+start+train_sample_count) % sample_count];
                memcpy(temp_train_samples.ptr(i), samples.ptr(j), sample_size);
            }

            predict(temp_test_samples, temp_test_responses, 0);
            for( i = 0; i < test_sample_count; i++ )
            {
                float val = temp_test_responses.at<float>(i);
                j = sidx[(i+start+train_sample_count) % sample_count];
                if( is_classification )
                    error += (float)(val != responses.at<int>(j));
                else
                {
                    val -= responses.at<float>(j);
                    error += val*val;
                }
            }
        }
        if( min_error > error )
        {
            min_error   = error;
            best_params = params;
        }
    }

    params = best_params;
    return do_train( samples, responses );

Best regards,

chris