I am testing the robustness of a predictive model that I have previously established. I used to withdraw multiple random samples to see how robust the model is. Few days ago I came across bootstrap sampling technique, which is claimed to "to improve the stability and accuracy of machine learning algorithms". The technique withdraws the same sample more than one time and completely ignores other samples, so how would that increase the stability and accuracy of a model? I am a bit confused about the advantages of this method, can someone give a little idea about it please?
(1) I don't think you've given an accurate example of bootstrap validation
a) it does draw random sample with replacement, but I'm not sure it "ignores" anything b) it doesn't increase stability/accuracy of the model, but rather of the test error
(2) This might be a useful starting point: Steyerberg: Internal validation of predictive models: efficiency of some procedures for logistic regression analysis (J Clin Epidemiol. 2001 Aug;54(8):774-81)