AIExplainer

What is target network?

A copy of a neural network used to provide stable targets for training another network

In deep learning, a target network is a duplicate of a neural network that is used to provide stable targets for the training of another network. This helps to improve the stability and performance of the training process.

Think of a target network like a blueprint for a house. Just as a blueprint provides a stable plan for building a house, a target network provides a stable target for the training of another network.

Target networks are used in training autonomous vehicles to learn how to drive safely. The target network provides a stable target for the training of the vehicle's control system.

Target networks are used in deep reinforcement learning to provide stable targets for the training of an agent. They are updated periodically to track the changes in the main network.

One common misconception is that the target network is the same as the main network. However, the target network is a separate copy of the main network that is updated periodically.

The concept of target networks was introduced in the paper 'Human-level control through deep reinforcement learning' by Mnih et al. in 2015.

reference network stable network target Q-network

Three products for different needs — explore what’s relevant to you.