SVM: The Dual Formulation

[@mr.norris3840]

Can't believe RUclips can educate me better than my lecture on Machine Learning. Thanks for helping me in exam season :D

[@PS3benimeni]

nicely explained. Thanks :)

[@EnabRasool]

How is this discriminant value calculated, at the end

[@pritomroy2465]

Thanks madam. This was a lot elaborate and informative.

[@abhishekbalawan6817]

Suggest some source for further reading on Lagrange's duality.

[@rishabhnegi6941]

but where are the python code??

[@siyuanchen5428]

good explanation on SVM. Thanks.

[@pkittali]

please explain in simple manner

[@santoshpalaskar6176]

Its already explained well and in simpler manner

[@pkittali]

@@santoshpalaskar6176 For a newbie its difficult to understand. Basically a basics on constrained optimization would give more clarity.

[@santoshpalaskar6176]

@@pkittali For that separate 2-3 chapters will be needed

[@pkittali]

@@santoshpalaskar6176 okay

[@lakshsinghania]

do the homework b4 watching this, u will get it very easily for sure

[@RahulDable]

Thanks.

[@nidhijoshi2978]

why do we subtract the constraints from the function to be minimised i.e. (1/2)||w||^2 ?? As it is added in the lagranges formula !!

[@deepakkumarshukla]

it's because g(w) constraints are supposed to be less than zero in langrangian. So, if you work out the constraint for SVM to be less than zero, it shall be negative and when adding that negative term appears as subtraction in Lagrangian function.

[@kelvinwu9844]

It is because in the later summation term you can see -ayx , which is also -w, so overall is -0.5w