Skip to content
Back to skills

Learning Values Across Many Orders Of Magnitude

ASecurity

**arXiv ID:** 1602.07714 **Authors:** Hado van Hasselt, Arthur Guez, Matteo Hessel, Volodymyr Mnih, David Silver **Published:** 2016-02-24T21:14:52Z **Abstract:** Most learning algorithms are not invariant to the scale of the function that is being approximated. We propose to adaptively normalize the targets used in learning. This is useful in value-based reinforcement learning, where the magnitude of appropriate value approximations can change over time when we update the policy of behavior....

  • 3 stars
  • 0 votes
  • 0 copies
  • 7 views
  • Added September 11, 2026
toolsgoperformance

Works with

  • cli

Security analysis

A100/100

Scanned September 11, 2026

npx -y skills add hiyenwong/ai_collection --skill learning-values-across-many-orders-of-magnitude --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Learning Values Across Many Orders Of Magnitude?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Learning Values Across Many Orders Of Magnitude
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/hiyenwong-learning-values-across-many-orders-of-magnitude/badge)](https://www.skillsdirectory.com/skills/hiyenwong-learning-values-across-many-orders-of-magnitude)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
# Learning values across many orders of magnitude

**arXiv ID:** 1602.07714
**Authors:** Hado van Hasselt, Arthur Guez, Matteo Hessel, Volodymyr Mnih, David Silver
**Published:** 2016-02-24T21:14:52Z
**Abstract:**
Most learning algorithms are not invariant to the scale of the function that is being approximated. We propose to adaptively normalize the targets used in learning. This is useful in value-based reinforcement learning, where the magnitude of appropriate value approximations can change over time when we update the policy of behavior. Our main motivation is prior work on learning to play Atari games, where the rewards were all clipped to a predetermined range. This clipping facilitates learning across many different games with a single learning algorithm, but a clipped reward function can result in qualitatively different behavior. Using the adaptive normalization we can remove this domain-specific heuristic without diminishing overall performance.

## Skill Description

This skill is generated from the arXiv paper: Learning values across many orders of magnitude (1602.07714).

## How to Use

[To be filled in by the user or by future automation]

## References

- [arXiv:1602.07714](http://arxiv.org/abs/1602.07714v2)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

View all in tools →

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.