Research breakthrough · curated entry
Deep Q-Learning plays Atari from pixels
Human-level control through deep reinforcement learning
- Date
- February 26, 2015
- Category
- Research breakthrough
- Organization / Model
- DeepMind
- Evidence
- 1 source · 1 evidence family
Using one network architecture and one set of hyperparameters, DeepMind learned 49 Atari games from raw pixels and score signals, exceeding human test players in several. It joined deep networks with Q-learning and showed end-to-end reinforcement learning can handle high-dimensional perception.
Curator's note: Cross-checked against the Nature paper DOI.
Curated sources · 1
Cite this page
This page is a curated static snapshot and can be cited directly. Preferred format GB/T 7714—2025 (electronic resource [EB/OL]); the other 5 styles open below, together with the machine-readable files.
GB/T 7714—2025 (electronic resource [EB/OL])
AI 纪年. DQN 登上《Nature》，深度强化学习直接从像素学习控制策略[EB/OL]. (2015-02-26)[2026-11-03]. https://aihist.net/e/dqn-2015.html.
APA 7 · American Psychological Association
AI 纪年. (2015, February 26). DQN 登上《Nature》，深度强化学习直接从像素学习控制策略 [Web page]. https://aihist.net/e/dqn-2015.html
MLA 9 · Modern Language Association
"DQN 登上《Nature》，深度强化学习直接从像素学习控制策略." AI 纪年, 26 Feb. 2015, aihist.net/e/dqn-2015.html. Accessed 3 Nov. 2026.
Chicago 17 · notes-bibliography
AI 纪年. "DQN 登上《Nature》，深度强化学习直接从像素学习控制策略." Published Feb. 26, 2015. Accessed Nov. 3, 2026. https://aihist.net/e/dqn-2015.html.
IEEE · citation-reference list
"DQN 登上《Nature》，深度强化学习直接从像素学习控制策略," AI 纪年, 26 Feb. 2015. [Online]. Available: https://aihist.net/e/dqn-2015.html. [Accessed: 3 Nov. 2026].
ACM · author-date variant
AI 纪年. February 26, 2015. DQN 登上《Nature》，深度强化学习直接从像素学习控制策略. Retrieved November 3, 2026 from https://aihist.net/e/dqn-2015.html.
All six styles come from one template and can be compared word for word. [引用日期], Accessed, urldate and Y2 all carry this page's build date, 2026-11-03 — a static page is an immutable snapshot, so substitute the date you actually consulted it. The address in every style is the archive's canonical page: this English page translates the same snapshot, and the work keeps its name, so one record cites both. For one-click copying use the same entry in the timeline, which takes the day you opened it. Expand the other styles before printing.