如何绘制高斯朴素贝叶斯分类器的决策边界？

Question

我使用下面的玩具数据集（类成员变量＆amp; 2特征）来应用高斯朴素贝叶斯模型并绘制特定于类的双变量正态分布的轮廓。

如何为决策边界添加一条线到下面的图？ （比如这里：https://alliance.seas.upenn.edu/~cis520/dynamic/2016/wiki/uploads/Lectures/2class_gauss_NB.jpg）

# Packages
library(klaR)
library(MASS)

# Data
d <- structure(list(y = structure(c(1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 1L), .Label = c("0", "1"), class = "factor"), x1 = c(2, 2.8, 1.5, 2.1, 5.5, 8, 6.9, 8.5, 2.5, 7.7), x2 = c(1.5, 1.2, 1, 1, 4, 4.8, 4.5, 5.5, 2, 3.5)), .Names = c("y", "x1", "x2"), row.names = c(NA, -10L), class = "data.frame")

# Naive Bayes Model
mN <- NaiveBayes(y ~ x1+x2, data = d)

# Data

# Class 1
m1 <- mean(d[which(d$y==1),]$x1)
m2 <- mean(d[which(d$y==1),]$x2)
mu1_2 <- c(m1,m2)  # Mean
sd1 <- sd(d[which(d$y==1),]$x1)
sd2 <- sd(d[which(d$y==1),]$x2)
Sigma1_2 <- matrix(c(sd1, 0, 0, sd2), 2)  # Covariance matrix
bivn1_2 <- mvrnorm(5000, mu = mu1_2, Sigma = Sigma1_2 )  # from Mass package: Simulate bivariate normal PDF 
bivn1_2.kde <- kde2d(bivn1_2[,1], bivn1_2[,2], n = 50)   # from MASS package: Calculate kernel density estimate


# Class 0
m3 <- mean(d[which(d$y==0),]$x1)
m4 <- mean(d[which(d$y==0),]$x2)
mu3_4 <- c(m3,m4)  # Mean
sd3 <- sd(d[which(d$y==0),]$x1)
sd4 <- sd(d[which(d$y==0),]$x2)
Sigma3_4 <- matrix(c(sd3, 0, 0, sd4), 2)  # Covariance matrix
bivn3_4 <- mvrnorm(5000, mu = mu3_4, Sigma = Sigma3_4 )  # from Mass package: Simulate bivariate normal PDF 
bivn3_4.kde <- kde2d(bivn3_4[,1], bivn3_4[,2], n = 50)   # from MASS package: Calculate kernel density estimate


# Plot
plot(x= d$x1, y=d$x2, xlim=c(-1,10), ylim=c(-1,10), col=d$y, pch=19, cex=2, ylab="x2", xlab="x1")
contour(bivn1_2.kde, add = TRUE, col="darkgrey")     # from base graphics package
contour(bivn3_4.kde, add = TRUE, col="darkgrey") # from base graphics package
text(labels = "Class 1",x = 8, y=7, col="grey")
text(labels = "Class 0",x = 0, y=4, col="grey")