1
$\begingroup$

Is it possible to form a table that will have simply the shortest distance from each source to destination using q learning?

If not, suggest any other learning algorithm.

$\endgroup$
  • $\begingroup$ It is not clear what the learning environment is here, and I think that is important. For instance, if the sources and destinations are in the real world, without accurate mapping, and you must actually physically explore the environment, then RL might be a reasonable choice. If the sources and destinations are on some kind of computer grid or maze, then simpler search algorithms might be better. Could you give more details about the learning problem you are faced with? $\endgroup$ – Neil Slater Sep 2 '18 at 18:34
1
$\begingroup$

Welcome to AI.SE Adarsh!

This is fairly simple to do with Q-Learning.

If you assign a reward equal to -1 for each location other than the goals, and 0 for the goals, then Q-learning with no discount factor will learn the length of the shortest path from each location to the goal. Alternatively, you can use a discounting factor, and then invert the equation on the learned values of the Q function to obtain the number of steps.

Hope this helps!

$\endgroup$

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.