How does SVM behave when number of features >> number of samples?

SVM performs well in high feature low sample scenarios due to margin maximization.

In such cases, data is often linearly separable or near separable. SVM avoids overfitting by maximizing margin and using regularization, making it effective for genomics and text data.