1. Learn
    2. /
  2. Courses
    3. /
  3. Deep Reinforcement Learning in Python
Connected

Exercise

Hands-on with Optuna

Use Optuna to optimize the hyperparameters of a simple function.

In practice, you would want to optimize an objective function which is expensive or time-consuming to evaluate. As a result, you want to find reasonable hyperparameters in as few trials as possible.

For convenience, you will use a predefined objective function here instead which can be evaluated near-instantaneously:

$$f(x,y) = 2*(1-x)^2 + (y-x)^2$$

The metric() function is defined in your environment.

For this exercise, x and y are the hyperparameters that you optimize for.

Instructions 1/3

undefined XP
    1
    2
    3
  • Declare the hyperparameters x and y as floats allowed to vary uniformly between -5 and 5.