How to implement SVM (Support Vector Machine) from scratch with Python

[@MariamElseedawy--]

I am so lucky to find this channel!

[@WassupCarlton]

This was very crispy and straightforward. bravo my guy. Going to attempt my own implementation now (with yours as guidance/inspiration) and hope it doesn't explode in me face.

[@srisaisubramanyamdavanam9912]

Here he used primal problem to derive equations from scratch which is not efficient for non linear classifications(because we use kernels there and primal is computationally expensive).Ill agree dual is more complicated to derive but that is efficient

[@SalehGoodarzian]

Thanks for making such a great educational video. Can you please kindly tell me how you recorded the screen to make this video?

[@nickfast8624]

very helpful ans useful!

[@marco8673]

so basically is an enhacment of the perceptron algorithm right?

[@سعیداحمدی-ل1ث]

ممنون لطفا بیشتر ویدیو بسازید

[@126sivgucsivanshgupta2]

where and when would 1 use svm in the real world, is it just an alternative to other regressors and classifiers ?

[@user-wr4yl7tx3w]

great tutorial

[@AsilKhalifa]

Great!

[@saputrauta09]

still confused why u used w.x-b instead of w.x+b, can anyone help me?

[@ajmeryexperiences4186]

You can use both , this will not impact on optimising weights

[@saputrauta09]

@@ajmeryexperiences4186 thanks sir

[@sriramshivkumar1721]

the accuracy im getting is very can you please tell me how to increase the accuracy of the model

[@BradKoriya]

can PHP help to access vector database? Everyone is talking about Python, so just want to know that if PHP can help to implement machine learning project, kindly let me know your suggestion.

[@isseym8592]

I don't know how that works. Maybe u can, but you might have to calculate vector products manually which will be a pain and probs unrealistic since you would have to calculate multi dimensional vectors if you want any application in the real world.

[@gauravbargaway]

How we can implement SVM for multiclass ?

[@turquoisegem441]

Need to implement multiple SVMS. Try 1 label against the rest and so on...

[@abhishekkaddipudi2436]

Use one to rest technique for the same algorithm

[@zyadbrave9554]

Slides please