---
title: "Understanding deep learning requires rethinking generalization"
description: "The random-label control shows that a network can fit arbitrary training targets while performing near chance on held-out examples. Read it to distinguish fitting from generalization, understand what stays fixed in the experiment, and avoid overclaims about regularization or learning theory."
canonical_url: "https://fanout.sh/daily/2026-10-09-rethinking-generalization"
md_url: "https://fanout.sh/daily/2026-10-09-rethinking-generalization.md"
last_updated: "2026-10-09"
access: "public"
---

# Understanding deep learning requires rethinking generalization

The random-label control shows that a network can fit arbitrary training targets while performing near chance on held-out examples. Read it to distinguish fitting from generalization, understand what stays fixed in the experiment, and avoid overclaims about regularization or learning theory.

## Paper details

- Authors: Chiyuan Zhang et al.

- Venue: ICLR 2017

- Track: AI Research

- Difficulty: Medium

- Reading time: 32 min

- Original paper: https://arxiv.org/abs/1611.03530v2

## What you will learn

- Why random targets must stay fixed across epochs for the paper’s memorization control.

- How equal training accuracy can accompany very different held-out accuracy in a matched comparison.

- Why the results challenge coarse explanations without proving that regularization or learning theory is useless.

## Continue exploring

[Map generalization](https://fanout.sh/knowledge-graph?node=ai-research%3Aconcept%3Ageneralization): Connect the random-label control to generalization, regularization, and inductive bias in the Fanout knowledge map.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md