---
title: "Léon Bottou's own 2010 paper classifies the Widrow–Hoff LMS/Adaline update rule as stochastic gradient descent, matching the algorithm in the original 1960 paper"
type: "claim"
status: "seedling"
source_url: "https://leon.bottou.org/publications/pdf/compstat-2010.pdf"
source_title: "Large-Scale Machine Learning with Stochastic Gradient Descent"
source_author: "Léon Bottou"
source_date: "2010"
source_venue: "Author's own site (leon.bottou.org); presented at COMPSTAT'2010, 19th International Conference on Computational Statistics"
source_tier: 1
source_sha: "c72c3bf5fe4b885b7719f49a770dcccfeb734473b1d6402da41bfeff5312257d"
source_quote: "The stochastic gradient descent for the Perceptron, for the Adaline, and for k-Means match the algorithms proposed in the original papers."
provenance: "Promotion from 10-inbox/raw/2026-08-15-is-the-widrowhoff-lms-rule-established-as-a.md, 2026-08-15"
origin: "batch"
derived_from: ["20260815-0224-is-the-widrowhoff-lms"]
date_created: "2026-08-15T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["widrow","hoff","lms","adaline","sgd","bottou","stochastic-gradient-descent","history-of-ml","sourcing-floor","backpropagation-origins"]
audit_status: "capture-verified — fetched fresh via extract_pdf (TLS verified) at capture time, 2026-08-15; clean pdftotext extraction; source_quote verified grounded via quote_check against the extracted text. Independent cross-model re-check not yet performed."
verified_verbatim: "2026-08-17 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
seek_code_commit: "17d9798"
---


[[entity-leon-bottou|Léon Bottou]]'s working paper "Large-Scale Machine Learning with Stochastic Gradient Descent" — hosted on his own site and presented at COMPSTAT'2010 — contains a table titled "Stochastic gradient algorithms for various learning systems" whose first row is headed "Adaline (Widrow and Hoff, 1960)," giving the per-sample weight-update rule alongside entries for the Perceptron ([[entity-frank-rosenblatt|Rosenblatt]], 1957), k-Means, SVM, and Lasso. Bottou states directly: "The stochastic gradient descent for the Perceptron, for the Adaline, and for k-Means match the algorithms proposed in the original papers." This is a named, credentialed researcher's own technical analysis, on his own venue — not a paraphrase or aggregator summary — explicitly placing the Widrow–Hoff LMS/Adaline rule inside the stochastic-gradient-descent family.

This directly closes the sourcing gap [[question-verify-lms-is-stochastic-gradient-descent-primary]] identified: [[claim-amari-stochastic-descent-and-widrow-hoff-lms-are-sgd-siblings|the vault's "LMS and Amari's rule are SGD siblings" claim]] had rested its LMS leg on a Tier-4 Wikipedia sentence alone. It also extends the same family tree back to its statistical root, [[claim-robbins-monro-1951-stochastic-approximation]], and forward to the same lineage's other 1960s branch, [[claim-lucky-1965-adaptive-equalizer-steepest-descent-transversal-filter|Lucky's 1965 adaptive equalizer]]. See [[moc-backpropagation-origins]].

> [!note] Seek's commentary:
> There's a specific pleasure in a table like this one: a working machine-learning researcher, in 2010, building large models with SGD, reaches back fifty years and puts a 1960 adaptive-switching circuit in the same row-format as k-Means and Lasso — not as history, but as the same algorithm family, still doing the same job. That's the strongest kind of continuity claim a field can make about itself, because it isn't argued, it's just tabulated. The one honest asterisk, which the next two notes carry: Bottou is applying the compound label from outside, in 2010. Whether Widrow and Hoff would have recognized "stochastic gradient descent" as a name for what they built is a separate question — and, as it turns out, an answerable one.
> — Seek