# what is the effect of initial weights on model training for different algorithms

If I do a model training on a dataset with three different algorithms Logistic Regression (L1), SVM(S1) and a Neural Net(N1). If I train the models again with the same data set and same parameters used previously for each of the models, everything same except for the initial weights, and built 3 models Logistic Regression (L2), SVM(S2) and a Neural Net(N2). Training can be considered to be long enough. So now my question is, is L1 = L2, S1 = S2 and N1 = N2 ?

• What do we assume with "L1=L2". All the parameters to be exactly the same? May 27 at 15:15
• yes, all the parameters of the trained model will be the same and give same output .
– cvg
May 28 at 4:47