Vladimir Vapnik

Vladimir Naumovich Vapnik was born on December 6, 1936.⁠[1]

He was born in the former USSR.⁠[2]

In 1958 he received a master’s degree in mathematics from Uzbek State University in Samarkand, USSR.⁠[3]

From 1961 to 1990 he worked at the Institute of Control Sciences in Moscow, rising to head of its Computer Science Research Department; in 1964 he completed a PhD in statistics there.⁠[2]

The institute’s history records that in the early 1960s he and Alexey Chervonenkis joined Lerner’s Laboratory and from 1962 to 1971 developed the “generalized portrait” method for pattern recognition.⁠[1]

He and Chervonenkis obtained results on the uniform convergence of relative frequencies in draft form in July 1966, announced them in a 1968 note in Soviet Mathematics Doklady, and gave full proofs in a 1971 paper in Theory of Probability and Its Applications.⁠[4]

The paper gave sufficient conditions for uniform convergence of frequencies to probabilities over a class of events, and necessary and sufficient conditions taking the probability measure into account.⁠[5]

The IEEE History Center biography says the resulting Vapnik-Chervonenkis (VC) learning theory established a fundamental quantity representing the limitations of learning machines, and that he later created the principles of structural risk minimization to handle generalization.⁠[6]

In 1974 his Theory of Pattern Recognition, co-written with Chervonenkis, was published in Moscow.⁠[1]

In 1982 Springer published his Estimation of Dependences Based on Empirical Data in New York.⁠[7]

After moving to the United States in 1990 he joined AT&T Bell Labs, where he and colleagues developed the theory of support vector machines.⁠[8]

In 1992 he, Bernhard E. Boser and Isabelle Guyon presented at the Annual Conference on Computational Learning Theory (COLT) a training algorithm that maximizes the margin between training patterns and the decision boundary, with the solution expressed as a linear combination of the “supporting patterns” closest to the boundary.⁠[9]

In 1995 he and Corinna Cortes published “Support-Vector Networks” in the journal Machine Learning, extending support-vector networks to non-separable training data and comparing them with classical algorithms on an optical character recognition benchmark.⁠[10]

That year he was appointed Professor of Computer Science and Statistics at Royal Holloway, University of London.⁠[3]

His The Nature of Statistical Learning Theory, published that year, described the general conditions under which a computing or biological system can learn a task, and how fast.⁠[7]

In 2002 he joined NEC Laboratories in PrincetonPrinceton UniversityPrinceton University is a private university and Ivy League member in Princeton, New Jersey. Chartered on October 22, 1746 as the College of New Jersey, it moved to Princeton in 1756 and adopted its present name in 1896. Undergraduate education leads to A.B. and B.S.E. degrees and emphasizes small-group discussion, independent research and residential college life. Graduate education spans the humanities, social sciences, natural sciences and engineering. As of October 2026, Christopher L. Eisgruber was president. Its 2026–27 operating budget was $3.407 billion, including $342 million for undergraduate aid.Open the full entry, staying until 2014, and from 2003 he was also a professor of computer science at Columbia University.⁠[2]

In 2006 he was elected to the National Academy of Engineering for “insights into the fundamental complexities of learning and for inventing practical and widely applied machine-learning algorithms.”⁠[11]

In December 2011 the Franklin Institute announced its 2012 laureates, awarding him, then at NEC Laboratories, the Benjamin Franklin Medal in Computer and Cognitive Science for fundamental contributions to the understanding of machine learning and for inventing widely used machine learning techniques.⁠[12]

In November 2014 Facebook AI Research announced that he had joined, describing him as the father of statistical learning theory and co-inventor of the support vector machine.⁠[13]

At FAIR he joined long-time collaborators including Jason Weston, Ronan Collobert and Yann LeCun; in 1995 he had bet LeCun that by 2000 neural networks would no longer be used for anything serious, and he lost the bet.⁠[14]

In December 2016 Columbia University announced that he had been awarded the IEEE John von Neumann Medal for developing statistical learning theory and support vector machines; his earlier honors included the 2008 Paris Kanellakis Award.⁠[8]

On November 16, 2018, he appeared on episode 5 of the Lex FridmanLex FridmanLex Fridman is a Soviet-born, US-raised computer scientist and podcast host who works as an MIT research scientist on human-AI interaction and autonomous driving. His Lex Fridman Podcast is known for hours-long conversations with guests including heads of state and technology leaders.Open the full entry Podcast, “Statistical Learning,” introduced as co-inventor of support vector machines, support vector clustering and VC theory; the conversation covered the limits of current approaches and open problems in AI.⁠[15]

In 2019 he and Rauf Izmailov proposed in the journal Machine Learning the “Learning Using Statistical Invariants” (LUSI) paradigm, in which a learning machine computes problem-specific statistical invariants and then minimizes expected error while preserving them, using both strong and weak convergence mechanisms.⁠[16]

In 2020 he and Izmailov published a paper in the proceedings of the Ninth Symposium on Conformal and Probabilistic Prediction and Applications calling a theory that uses both strong and weak modes of convergence the “complete statistical theory of learning,” and describing algorithms including LUSI-SVM and LUSI-NN.⁠[17]

On February 14, 2020, he returned to the Lex Fridman Podcast for episode 71, “Predicates, Invariants, and the Essence of Intelligence,” discussing predicates, strong and weak convergence, deep learning and the essence of intelligence, symbolic AI and the role of philosophy in learning theory.⁠[18]

In February 2020 the BBVA Foundation announced that its 12th Frontiers of Knowledge Award in Information and Communication Technologies would go to him, Isabelle Guyon and Bernhard Schölkopf for fundamental contributions to machine learning; the jury called support vector machines and kernel methods “a major machine learning paradigm in both research and applications.”⁠[19]

After learning of the award he said: “You cannot give the machine the rule, the machine has to learn the rule.”⁠[19]

Columbia University: Official website: https://www.columbia.edu/ (opens in a new window)⁠[8]

Institute of Control Sciences: Official website: https://www.ipu.ru/en (opens in a new window)⁠[3]