search

IntelLabs / coach

Reinforcement Learning Coach by Intel AI Lab enables easy experimentation with state of the art Reinforcement Learning algorithms
https://intellabs.github.io/coach/
Apache License 2.0
2.32k stars 461 forks source link

Coach Designer for Visual programming creation of RL models by Drag-and-Drop #317

Closed developeralgo8888 closed 5 years ago

developeralgo8888 commented 5 years ago

Feature request . its very daunting task to learn several Machine Learning libraries every 2 to 3 months for Data Scientists or new users . Is Intel working on a " COACH DESIGNER " similar to Microsoft's Azure Machine Learning Studio which allows for Drag-and-Drop visual programming and then export the Code so that you can run the complete Model python script locally or externally. That would be a " GOD sent " Feature to anyone . Its exhausting for Data Scientists , Geeez !!!!!!

gal-leibovich commented 5 years ago

Hi @developeralgo8888, thanks a lot for the suggestion. It really would be a blessing if we could get reinforcement learning to work so easily with just a drag-and-drop interface. Unfortunately, this is not currently the case with Reinforcement Learning. Getting RL algorithms to work well on a new problem is often not an easy task. The state-of-the-art in RL is still not a push of a button getting it solve a new problem. It often requires a deeper understanding of the algorithm and the problem, in order to get it to work right.

I don't necessarily agree with all of it, but there are some great parts in this blog post, which describes some of the current issues with RL today.

Thanks a lot for the idea, but we will not be pursuing this direction at the moment.