# Double DQN

Separates action selection from action evaluation to correct the overestimation bias of plain DQN.

## What it is used for

The same discrete-action problems as DQN, with more stable value estimates.