We can, in a very similar way, implement logistic regression as an R6 class. The code is included just for intellectual curiosity, as it is closely related. 

There is not a lot of difference between the perceptron and logistic regression. They share a lot in common, the main difference being the activation function (logit instead of the Heaviside step function), which also changes the update rule for the weights. For convenience, we highlight in bold the more relevant differences:

library(R6)
logit <- function(x){
 1/(1+exp(-x))
}
LR <- R6Class("LR", 
     public = list(
     dim = NULL,
     n_iter = NULL,
     learning_rate = NULL,
     w = NULL,
     initialize = function(learning_rate = 0.25, n_iter=100, dim=2){
         self$n_iter <- n_iter
         self$learning_rate <- learning_rate
         self$dim <- dim
         self$w <- matrix(runif(self$dim+1), ncol = self$dim+1)
 }
 , forward = function(x){
         dot_product <- sum(x*self$w)
         y <- logit(dot_product)
         return(y)
     }
 , backward = function(t,y,x){
         for(j in 1:ncol(x)){
         self$w[j] <- self$w[j]+self$learning_rate*(t-y)*x[j]*logit(x[j])*(1-logit(x[j]))
 }
 
 }
 , train = function(X,t){
         X <- cbind(-1,X) #add bias term
         n_examples <- nrow(X)
 
 for(iter in 1:self$n_iter){
     for(i in 1:nrow(X)){
     y_i <- self$forward(X[i,])
     self$backward(t[i],y_i, X[i,])
 }
 if(iter %% 20 == 0){
     cat("Iteration: ", iter)
     print("Weights: ")
     print(unlist(self$w)) 
 }
 
 }
 }
 , predict = function(X){
     X <- cbind(-1,X) #add bias
     preds <- c()
     for(i in 1:nrow(X)){
         preds[i] <- self$forward(X[i,])
     }
 return(preds)
 }
 )
)

As we can see, there are not a lot of changes with respect to the previous code, the main action happening on the backward step.