在给定高斯过程模型的情况下计算函数的似然性

Question

我使用scikit-learn拟合高斯过程回归。 （我的实际上是一个简单的一维案例）

#include <stdio.h>

struct rettype {
    int start;
    int end;
    int sum;
};

struct rettype max_subarray(int array[], int length);

int main(void){
    int array[20];
    int len = 0, num;
    struct rettype ret;

    printf("Type length in.\n");
    if(scanf("%d", &len) != 1){
        printf("invalid input.\n");
        return 1;
    }
    if(len > 20){
        printf("too long.\n");
        return 2;
    }
    for(num = 0; num < len; num++){
        printf("%dth element: ", num);
        if(scanf("%d", &array[num]) != 1){
            printf("invalid input.\n");
            len = num;
            break;
        }
    }
    if(len > 0){
        ret = max_subarray(array, len);
        printf("Maximum Subarray:\nfrom element %d to element %d with sum of %d\n", ret.start, ret.end, ret.sum);
    }
    return 0;
}

struct rettype max_subarray(int array[], int length){
    struct rettype ret  = { .sum = array[0] };
    struct rettype curr = { .sum = array[0] };

    for(int i = 1; i < length; ++i){
        if(array[i] < array[i] + curr.sum){
            curr.sum += array[i];
            curr.end = i;
        } else {
            curr.sum = array[i];
            curr.start = curr.end = i;
        }
        if(ret.sum < curr.sum){
            ret = curr;
        }
    }
    return ret;
}

一旦这个模型适合，我想计算另一个数据集（from sklearn.gaussian_process import GaussianProcessRegressor from sklearn.gaussian_process.kernels import RBF, WhiteKernel import numpy as np # Some example data x = np.linspace(0,10,100) y = np.sin(x) + np.random.normal(0, 0.1, x.shape) # Fitting the model some_kernel = RBF() + WhiteKernel() gpr = GaussianProcessRegressor(kernel=some_kernel) gpr.fit(x[:, None], y) ）在这个拟合模型下的概率。例如y1。 一种非常天真的方式。

P(y1|GP_fit_on_y)

但这不会考虑GP的协方差结构。关于如何做的任何想法？