📊

Vladimir Vapnik

Original nameVladimir Vapnik

Soviet-American Mathematician, Founder of Statistical Learning Theory and the Support Vector Machine

The Math Behind the Algorithm · Theoretical Founders

Statistical Learning TheoryThe Support Vector Machine (SVM)VC Dimension Theory

Who they are

Vladimir Vapnik (1936– ) is a mathematician born in the Soviet Union and later emigrated to the US who, with Alexey Chervonenkis in the Soviet period, co-proposed the "VC dimension," a theoretical framework measuring the relation of model complexity and generalization ability, giving a rigorous mathematical explanation for the fundamental question of machine learning "why a model can learn from limited samples a rule that generalizes to new data." Vapnik thereafter at Bell Labs further proposed the "support vector machine" (SVM) algorithm, which in the decade-plus from the 1990s to the early 2000s was the mainstream first choice for the great majority of classification and regression machine-learning tasks, its rigorous, solid theoretical basis and relatively small data requirement making it, for a long time before the rise of deep learning, seen as a more reliable and more scholarly-favored technical path than early neural networks.

Primary sourcesVapnik, The Nature of Statistical Learning Theory (1995)

Key stories

Before the Rise of Deep Learning, SVM Was the "Standard Answer"

Before AlexNet proved deep learning’s overwhelming advantage in 2012, the support vector machine, by the rigorous statistical-learning-theory basis Vapnik proposed, was for over a decade the mainstream first choice of academia and industry for classification machine-learning tasks, many researchers even for a time holding the neural-network path proven a dead end lacking theoretical guarantee, and this history also shows, from the side, that the mainstream standing of deep learning taken for granted today was in fact established only relatively recently in the history of AI.

Relationships

Ally

🕸️Judea Pearl

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

The "standard answer" before the rise of deep learningThe support vector machine was for over a decade the mainstream first choice for classification tasks, a history often cited to show that the mainstream standing of a technical path is not fixed.

Appears in

The Perceptron, the Winter, and the Revival of ConnectionismStatistical Learning Theory and Bayesian Networks

← Back to Figures