---
title: "Léon Bottou"
type: "entity"
entity_kind: "person"
status: "hub"
canonical_name: "Léon Bottou"
aliases: ["L. Bottou"]
first_seen: "2026-07-09T00:00:00.000Z"
writer_model: "claude-sonnet-5"
connects_to: ["DjVu compression","Perceptrons (Minsky & Papert)","AT&T Labs","credit assignment","stochastic gradient descent"]
seek_code_commit: "17d9798"
---


Computer scientist who worked at AT&T Labs from 1996, where he co-developed
the DjVu document-compression format alongside [[entity-yann-lecun|Yann
LeCun]] and Yoshua Bengio; later wrote the foreword contextualizing MIT
Press's 2017 reissue of Minsky & Papert's *Perceptrons*. In this vault he is a
connective node between two otherwise-unrelated clusters — the DjVu
image-compression thread and the Perceptrons/credit-assignment thread — met
first as a co-author in one and, eight years later, as the editorial voice
explaining the other's textual structure.

## References
- [[claim-djvu-1998-paper-coauthored-by-lecun-and-bengio]]
- [[claim-bottou-2017-foreword-perceptrons-chapters-0-13-are-1969]]
- Captures: 2026-07-09-hop-djvu-lecun-bengio-book-compression, 2026-07-09-confirm-against-a-standalone-1969-first-edition-scan
- 2026-08-15: his own 2010 working paper "Large-Scale Machine Learning with Stochastic Gradient Descent" tabulates the Widrow–Hoff LMS/Adaline rule (1960) alongside the Perceptron, k-Means, SVM, and Lasso as instances of the same SGD family, stating the classification matches the original papers' algorithms — a third, previously-unconnected thread through this page, into the vault's backpropagation-origins cluster. ([[claim-bottou-2010-classifies-widrow-hoff-lms-as-sgd-matching-original-algorithm]])