Author(s): Ananya

Originally published on Towards AI.

I write articles on Data Science, Finance and philosophy. In this one, I am focusing on one of the popular machine learning algorithms SVM, if you are someone who like reading about these topics feel free to subscribe. These question are more of conceptual based and interview centric.

PS: I like putting doodle images as illustrations, it helps me remember better and also keeps things interesting.

I have covered 15 questions in Part 1 and will cover the 10 more questions in this part. You can check out Part 1 here.

1. Why do linear classifiers fail on non-linearly separable datasets?

2. How can feature transformation techniques help solve non-linear classification problems?

3. What is the kernel trick in Support Vector Machines?

4. Why are kernel functions considered powerful in SVM models?

5. Compare linear, polynomial, and RBF kernels used in SVMs.

6. What are the disadvantages or challenges of using kernel methods?

7. Is SVM inherently a linear algorithm or a non-linear one? Justify your answer.

8. How can SVMs be extended to handle multi-class classification tasks?

9. Are kernels exclusive to SVMs, or are they used in other algorithms as well?

10. Under what circumstances would SVM perform better than Logistic Regression?

Published via Towards AI