Reinforcement Learning Prediction Algorithms

cjjc20001cjjc20001 Member Posts: 8 Contributor II
edited December 2018 in Help

Dear All,

 

I am relatively a newcomer in Rapidminer but so far it has been very useful. I am doing a research on vulnerability prediction. My question is that while Rapidminer is equipped with a lot of production models, I can't seem to find an operator that related to specfically reinforcement learning such as Deep Q. Am I missing something or there is really no tool yet available in Rapidminer?

I would appreciate your help on this matter as I am interested in machine learning techniques.

 

 

Tagged:

Answers

  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,635 Unicorn

    Not really.  Reinforcement learning is typically deployed in systems where the computer is trying to achieve some predetermined outcome and is modifying inputs and parameters subject to some constraints in order to do that, and theoretically improving in each successive iteration as it adjusts these inputs.  So in principle it has a lot of overlap with supervised machine learning problems, but the timeframe of the application is different, because in classic machine learning, you already have all the of labeled outcomes available at the time you are constructing the model.  

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
  • cjjc20001cjjc20001 Member Posts: 8 Contributor II

    Thank you for the reply. Correct if I am wrong, so basically there is no direct tool in reinforcement learning in Rapid MIner. 

     

  • Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,761 Unicorn

    @cjjc20001 RapidMiner has the Keras extension which allows you to use TensorFlow, CNTK, and Theano. From there you can build RNN, LSTM, and other types of deeplearning processes. I assume when you refer to reinforcement learning algorithms you mean deep learning?

  • cjjc20001cjjc20001 Member Posts: 8 Contributor II

    I get what you are saying. But I was looking for something like what you already have in Rapidminer. It would seem that the suggestions you gave me would require me to reconfigure those tools in order to make them work for my prediction models. 

  • pschlunderpschlunder Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, RMResearcher, Member Posts: 96 RM Research

    Hi @cjjc20001,

    you are right, RapidMiner does not support Reinforcement Learning directly, yet.

     

    Regards,

    Philipp

     
  • BoryzoBoryzo Member Posts: 1 Contributor I
    If Rapidminer does not support RL(Reinforcement Learning) directly, What is the method of how to use 'Q-learning' indirectly?
Sign In or Register to comment.