How does SVM behave in ultra-high dimensional sparse spaces?

SVM performs well due to margin-based learning and sparsity exploitation.

In ultra-high dimensions like NLP, most features are zero. SVM efficiently finds separating hyperplanes using only informative support vectors.