3
$\begingroup$

I have an agent (drone) that has to allocate subchannels for different types of User Equipment.

I have represented the subchannel allocation with a 2-dimentional binary matrix, that is initialized to all zeros as there is no requests at the beginning of the episode.

When the agent chooses an action, it has to choose which subchannels to allocate to which UEs, hence populating the matrix with 1s.

I have no idea how to do it.

$\endgroup$
  • $\begingroup$ Can you please be a little more intuitive on what you would like to do? What is a subchannel, UE, and in general what a human being would face in this problem. Describe a use case scenario. This would help to define our env, space and actions. $\endgroup$ – ddaedalus Sep 28 '20 at 9:47
0
$\begingroup$

I think this question is hinting at the problem of choosing an exploration strategy.

The simplest strategy is to use the so called epsilon-greedy strategy (or $\epsilon$-greedy). This means that you select an action at random $x$ percent of the times that an agent has to select an action. The other times, the agent takes the action that its current policy dictates. Usually, $x$ declines throughout the learning process.

There are many different exploration strategies to choose from, with the aforementioned one being really the most simple one. For getting an impression of how diverse these different approaches to exploration strategies are, please be referred to Google Scholar (searching for "reinforcement learning exploration strategies").

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.