Vladimir Vapnik

Co-creator, with Alexey Chervonenkis, of the VC (Vapnik-Chervonenkis) dimension — the measure of a learning model's capacity that founded statistical learning theory — and later of the support-vector machine.

Matters to this vault as the theoretical ancestor Yaser Abu-Mostafa's own foundational learning-theory papers name explicitly: across his 1989 VC-dimension paper, his 1992/93 NIPS hints paper, and his 1995 "Hints" review, the one citation recurring in all three as the stated ancestry of the VC dimension is Vapnik and Chervonenkis (1971), "On the uniform convergence of relative frequencies of events to their probabilities" (Theory of Probability and its Applications 16, 264–280) — credited in the 1995 review jointly with Blumer et al. (1989) rather than singularly, and in none of the three alongside an optimal-control or LQG source. That absence is the direct evidence behind question-verify-abu-mostafa-learning-theory-cites-optimal-control-lqg, and places Vapnik, not any control-theory lineage, as the actual intellectual root of the NIPS-founder-adjacent theory this vault checked (observation-nips-founders-kwakernaak-twente-founding-cosine-pairing-is-embedding-false-friend).

Chervonenkis has no hub of his own yet — carried here as a mention, since the vault's current footprint on the pair rests entirely on their joint 1971 paper rather than on either man's separate later work.

References

• claim-abu-mostafa-1995-hints-review-cites-no-optimal-control-lqg-source — the 1971 paper as one of the 1995 review's two jointly stated theoretical ancestors, with Blumer et al. (1989)
• claim-abu-mostafa-1989-and-1993-papers-also-cite-no-optimal-control-lqg-source — the same citation recurring across two earlier papers
• observation-nips-founders-kwakernaak-twente-founding-cosine-pairing-is-embedding-false-friend · entity-yaser-abu-mostafa

Updates

• 2026-09-29 (propagation-repair): the body sentence above was corrected in place — was: across the 1989, 1992/93 and 1995 papers "the one recurring, singular citation for 'the key tools for dealing with the information issue' is Vapnik and Chervonenkis (1971)," named "explicitly and exclusively"; now: the 1971 paper is the citation recurring in all three as the VC dimension's stated ancestry, credited in the 1995 review jointly with Blumer et al. (1989) rather than singularly, and the "key tools" phrase is dropped from this page because it belongs to Abu-Mostafa's 1989 paper, not the 1995 review. Per the 2026-09-29 cross-model audit of claim-abu-mostafa-1995-hints-review-cites-no-optimal-control-lqg-source, which re-read the 1995 review end to end and found it naming "The VC dimension (Blumer et al. 1989; Vapnik and Chervonenkis 1971)." The References line for that claim was corrected from "sole stated theoretical ancestor" to the joint framing for the same reason. The core negative — no optimal-control or LQG source in any of the three reference lists — is unaffected, as is Vapnik's standing here. The provenance and audit_status frontmatter keep their promotion-time wording ("the singular cited theoretical ancestor") as a dated historical record and were deliberately not rewritten.