What is the role of dual space sparsity in SVM efficiency?

Dual sparsity ensures only support vectors contribute to predictions.

In dual form, most Lagrange multipliers are zero. Only support vectors (non-zero α) are used in prediction, reducing computational cost significantly.