avivt
Why not just use Python Packages -Nobody uses the old grandfather called MATLAB
It is waste of time to learn 2 libraries especially damn MATLAB. I have not used MATLAB since i left University 6 years ago and never come across any company in real world using MATLAB . In some die-hard colleges that don't want to let go still use MATLAB but majority have moved on to python . There are Tensorflow , Theano and others that would accomplish the same task probably in easier way and help majority of people adopt your research
Well, my university training focused more on theory, and Matlab was the main tool, so when I started working on this project I coded some of it in Matlab. For research, Matlab's IDE still offers some advantages for quickly analyzing results, in my opinion.
But I agree, it would be great to convert the Matlab code to python. For an experienced python programmer this shouldn't take much time, and if you're interested in taking this on I'd be glad to assist. At present, I don't have the bandwidth to take this on myself.
thanks Avivt, i will try to migrate it to tensorflow /theano and python
Time is not on myside at the moment. i wish you could have written it in tensorFlow since you understand the experiments and your research better it would probably take you less than 3 hours to convert the entire code to tensorflow. Then we can further push your research and adopt it more. If i get a chance after the christmas holidays , i will give it a short
Seems like VIN are implemented here using Python:
https://github.com/TheAbhiKumar/tensorflow-value-iteration-networks
Actually i tested it but its not stable at all. it works only for Gridworld-8 but not on image size 16 and 28 . But even with Gridworld-8 it is not stable at all
@avivt @developeralgo8888
I have a full python + pytorch + numpy implementation of the gridworld experiments. I am happy to PR that to a new "python" branch for this repository or keep the release in my own repository. Let me know which you would prefer.
Thanks, Kent
Hi Kent, This is really great!! I think it's best to keep the code in your own repo, and I'll add a link to it on the info page.