How can I update an already trained machine learning model with only the new examples, without training the model over the whole dataset again?is it possible?

[CaoThien96]

First :
I train with var xor = [
[[0, 0], 0],
[[0, 1], 1],];
Then i save model
When have new data
I load and restore model
And train it with new data
var xor2=[
[[1, 0], 1],
[[1, 1], 0] ]
When i test with
var test = [
[[0, 0], 0],
[[0, 1], 1],
[[1, 0], 1],
[[1, 1], 0]
];
when result:
0 XOR 0 => 1 false
0 XOR 1 => 0 false
1 XOR 0 => 1 true
1 XOR 1 => 0 true

[CaoThien96]

Sorry for my bad English