Hill Climbing

## Gradient Ascent

Gradient ascent is similar to gradient descent.

• Gradient descent steps in the direction opposite the gradient, since it wants to minimize a function.
• Gradient ascent is otherwise identical, except we step in the direction of the gradient, to reach the maximum.

While we won't cover gradient-based methods in this lesson, you'll explore them later in the course!

## Local Minima

In the video above, you learned that hill climbing is a relatively simple algorithm that the agent can use to gradually improve the weights \theta in its policy network while interacting with the environment.

Note, however, that it's not guaranteed to always yield the weights of the optimal policy. This is because we can easily get stuck in a local maximum. In this lesson, you'll learn about some policy-based methods that are less prone to this.

## Additional Note

Note that hill climbing is not just for reinforcement learning! It is a general optimization method that is used to find the maximum of a function.