I am relatively new to reinforcement learning and have been experiencing with a reinforcement learning model to make decisions based on human activities (dynamic environment). Appreciate if someone can help me in understanding how best to evaluate a reinforcement learning model for performance before the model goes into production and when it is in production.

  • 1
    $\begingroup$ It will depend a lot on your problem and environment. Could you please give some constraints and details of the environment, what data you have from the real environment, whether you have a simulator for the environment, how you trained the agent (given it was not learning in a real environment), and what kind of algorithm was used - e.g. a Q-learning model like DQN, or a policy gradient model like A3C. $\endgroup$ – Neil Slater May 6 '19 at 7:42
  • $\begingroup$ The data is in the form of images (sensor data) representing the activities in a smart home like turn off the light, turn off the TV .etc.I have not trained the agent yet.What i want to do is to make decisions according to the situation in a dynamic environment.Regarding algorithm,i am planning to use Q-learning model like DQN. $\endgroup$ – Case Msee May 7 '19 at 0:25
  • $\begingroup$ So in your data, is something taking the same kind of decision as you want your bot to take? Or is it just some sensor data of different activities, with no actions taken by any agent, and no way to assign reward for correct/incorrect action? $\endgroup$ – Neil Slater May 7 '19 at 6:44
  • $\begingroup$ Alternatively, are the observations just actions (turn on light, turn off TV) without consequences? Or do you know when the observed action was correct or incorrect in each scenario (or had some other measurable consequence which you want to optimise)? $\endgroup$ – Neil Slater May 7 '19 at 7:00
  • $\begingroup$ Actually,human activities are monitered using video cameras and data is recorded.Now i want to convert the video records into image frames to pass thross CNN for feature extraction.Now the extracted features contain useful information of human activities like (sleeping,washing,Preparing the meal .etc).I want to pass these features to reinforcement learning to make continuous decisions in new situations.For more information please refer to this paper: lrec-conf.org/proceedings/lrec2014/pdf/118_Paper.pdf $\endgroup$ – Case Msee May 7 '19 at 11:51

Your Answer

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

Browse other questions tagged or ask your own question.