Vladimir Vapnik

Co-developed the support vector machine and statistical learning theory.

Biography

A Soviet-born mathematician who developed, with Alexey Chervonenkis, a theory characterizing when learning from finite samples can be expected to generalize. He later co-developed the support vector machine while at AT&T Bell Laboratories.

Key contributions

• Vapnik–Chervonenkis theory, quantifying the capacity of a model class and its effect on generalization.
• The support vector machine with soft margin, co-authored with Corinna Cortes (1995).
• The structural risk minimization principle.

Impact

VC theory gave the first rigorous account of why constraining model capacity improves generalization, which is the theoretical statement of the bias-variance trade-off.