---
title: "Yaser Abu-Mostafa"
type: "entity"
entity_kind: "person"
status: "hub"
canonical_name: "Yaser S. Abu-Mostafa"
aliases: ["Yaser Abu-Mostafa","Y. S. Abu-Mostafa","Yaser S. Abu-Mostafa"]
first_seen: "2026-07-12T00:00:00.000Z"
writer_model: "warden/claude-opus-4.8"
connects_to: ["NIPS/NeurIPS (founding)","machine learning theory","Ed Posner (co-founding general chairman)","Terry Sejnowski (the misattributed founder, defined-against)","history of machine learning"]
provenance: "Warden pass 2026-09-26 (warden/claude-opus-4.8), run per 00-meta/specs/seek-warden-spec.md on a maintenance engine kept separate from the notes' writers. Builds the Abu-Mostafa person hub the 2026-09-26 [entity] flag (seek-flags.md L5722) asked for (paired with Ed Posner — the Posner half is carried, not built: his vault identity beyond the founding role rests on unverified-synthesis and collides in name with the existing hub [[entity-richard-a-posner]]). Person test run honestly against a direct read this pass of the anchor claim-note (Tier 1, verified-verbatim from Abu-Mostafa's own Caltech page) and the false-friend observation that places the founding in context; not the flag summary. The writer's flag leaned 'likely a single-cluster decline', but single-cluster is a bar for orgs, not persons, and founding the domain's flagship conference is unambiguously load-bearing with a read primary. Build 2 of the run's <=2. Report: warden-2026-09-26.md."
audit_status: "2026-09-26 (warden/claude-opus-4.8, writer): single-model build on a maintenance engine; not yet cross-model audited. The founding-role fact is Tier 1, verified-verbatim (seek_verify, 2026-08-07, against a direct fetch of his Caltech page). The vault's footprint on him is currently narrow (one substantive claim plus one false-friend observation); the hub is built on the strength of the fact's centrality and its verified primary, not on breadth, and is honestly reversible to a mention if that never widens."
seek_code_commit: "f8a723c"
---


Yaser S. Abu-Mostafa is a Caltech professor of electrical engineering and computer science who was the founding program chairman of the NIPS/NeurIPS conference in 1987, alongside Ed Posner as founding general chairman.

Matters to this vault as the corrective to the more economical, wrong story of how the field's flagship venue began. It is tempting to assume the recurring connectionism name — Terry Sejnowski, author of NETtalk — founded the conference whose proceedings [[entity-yann-lecun|Yann LeCun]] later freely released; in fact Sejnowski was an invited first-year speaker who went on to organize NIPS 1988, while Posner and Abu-Mostafa founded it ([[claim-nips-1987-founders-posner-abu-mostafa-not-sejnowski]]). Abu-Mostafa's own Caltech page is the vault's source: "the founding program chairman of NIPS/NeurIPS in 1987, with the late Ed Posner as the founding general chairman." The founding event also sits, one field away, near the control-theory lineage the vault tracks: whether Abu-Mostafa's own learning-theory work cites any optimal-control/LQG source is the open check routed at [[question-verify-abu-mostafa-learning-theory-cites-optimal-control-lqg]] ([[observation-nips-founders-kwakernaak-twente-founding-cosine-pairing-is-embedding-false-friend]]).

What the vault has actually read of Abu-Mostafa: his own Caltech faculty page stating the 1987 founding roles (Tier 1, verified-verbatim). His broader learning-theory work is named but not yet read at primary here. Note a name-collision hazard: **Ed Posner**, his co-founder, is a different person from the judge-economist [[entity-richard-a-posner|Richard A. Posner]]; Ed Posner has no hub yet and is carried as a mention, not built, this pass.

## References
- [[claim-nips-1987-founders-posner-abu-mostafa-not-sejnowski]] — the founding attribution, from his own Caltech page (Tier 1, verified-verbatim); the correction against the Sejnowski assumption
- [[observation-nips-founders-kwakernaak-twente-founding-cosine-pairing-is-embedding-false-friend]] — the founding event as one seed of an embedding false friend, with the real field-level lineage running one level up through the 1973 disciplinary wall
- See also [[claim-nettalk-1987-complex-systems-distinct-from-1986-report]] (Sejnowski's NETtalk), [[claim-djvu-created-to-freely-distribute-scanned-documents]] (the freed NIPS archive), [[moc-backpropagation-origins]]

## Updates
- 2026-09-27: the open founder-level-debt check named above is now answered from the strongest available angle — his three core VC-dimension/hints theory papers (1989, 1992/93, 1995), read in full at the reference-list level, cite no optimal-control or LQG source; his stated theoretical ancestor throughout is Vapnik and Chervonenkis (1971) instead. His 2012 textbook *Learning From Data* remains unread and is the one gap left. ([[claim-abu-mostafa-1995-hints-review-cites-no-optimal-control-lqg-source]], [[claim-abu-mostafa-1989-and-1993-papers-also-cite-no-optimal-control-lqg-source]])