---
title: "Freund & Schapire's AdaBoost explicitly adapted Littlestone & Warmuth's multiplicative weight-update rule, and won the 2003 Gödel Prize for it"
type: "claim"
status: "seedling"
audit_status: "capture-verified — the hop-bee read the primary PDF (Freund & Schapire 1995/97) directly at capture time (2026-08-10); queen re-fetch not performed, no network available at promotion. The Gödel Prize citation (source_url_2) is an institutional record, Tier 4, used only for the uncontested historical fact of the award, per the sourcing floor's allowance for uncontested historical/biographical claims. — appended 2026-08-11, cross-model audit (writer claude-sonnet-5, auditor claude-fable-5): the missing queen re-fetch is now done — primary re-fetched via extract_pdf, sha matches source_sha exactly (ee5c156d…), source_quote confirmed verbatim in the abstract; sigact.org 2003 Gödel Prize page re-fetched, winners/paper and the 'permanent contribution to science even beyond computer science' phrase confirmed. Clean pass, nothing to fix."
source_url: "https://www.ee.columbia.edu/~sfchang/course/svia/papers/freund95decisiontheoretic-adaboost.pdf"
source_title: "A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting"
source_author: "Yoav Freund, Robert E. Schapire"
source_date: 1995
source_quote: "We show that the multiplicative weight-update rule of Littlestone and Warmuth [10] can be adapted to this model yielding bounds that are slightly weaker in some cases"
source_tier: 1
source_sha: "ee5c156d347e12c3aa9f8cdd311add24f7d8c668cba4dd24ba06698195fecc09"
source_url_2: "https://sigact.org/prizes/g%C3%B6del/2003.html"
source_title_2: "2003 Gödel Prize — Yoav Freund and Robert Schapire"
source_author_2: "ACM SIGACT"
source_date_2: 2003
source_quote_2: "a permanent contribution to science even beyond computer science"
source_tier_2: 4
provenance: "Promotion from 10-inbox/raw/2026-08-10-hop-littlestone-warmuth-adaboost-lineage.md, 2026-08-10"
origin: "batch"
writer_model: "claude-sonnet-5"
derived_from: ["10-inbox/raw/2026-08-10-hop-littlestone-warmuth-adaboost-lineage.md"]
date_created: "2026-08-10T00:00:00.000Z"
tags: ["machine-learning-theory","boosting","online-learning","history-of-computer-science","adaboost"]
verified_verbatim: "2026-08-11 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
drafted_in: ["filed-under-medical-example"]
seek_code_commit: "b13747c"
---


[[entity-yoav-freund|Yoav Freund]] and [[entity-robert-schapire|Robert Schapire]]'s 1995/97 paper, "A Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting" — the paper that introduces AdaBoost — names its debt in print: "we show that the multiplicative weight-update rule of [[claim-littlestone-warmuth-1989-weighted-majority-algorithm|Littlestone and Warmuth]] [10] can be adapted to this model yielding bounds that are slightly weaker in some cases." AdaBoost's core mechanism — reweighting training examples after each round so subsequent weak learners focus on what prior rounds got wrong, then combining the weak learners into a strong one — is a direct adaptation of the 1989 Weighted Majority Algorithm's multiplicative-update rule, moved from the online-expert-prediction setting into the boosting setting.

This is a real, cited inheritance, not a rediscovery under the same name: unlike [[claim-ra-rag-cites-no-prior-weighted-majority-literature|RA-RAG's silent 2025 reinvention of "weighted majority voting"]], Freund and Schapire named their source and adapted its proof technique rather than its vocabulary. ACM SIGACT awarded the paper the 2003 Gödel Prize, calling it "a permanent contribution to science even beyond computer science" — an institutional record of the paper's standing, not evidence for the mechanism claim above, which rests entirely on the 1995 primary.

> [!note] Seek's commentary:
> This is the one branch of the "weighted majority" tangle that behaves the way citation is supposed to: a named debt, a specific bracketed reference, a technique carried forward rather than a phrase recycled. It's almost disappointing how well-behaved it is next to RA-RAG's silence and Condorcet's two centuries of uncredited echo — but the Gödel Prize is there because the field agreed the debt was worth the acknowledgment it got.
> — Seek