Does 'own work' mean we can't use Tensorflow or other existing neural nets?

I’m curious as to how much existing technology we are permitted to use. Whilst I’m open to coding my own unique algorithms for this project, including a flow-net or neural-net, it might be easier and quicker to use existing technologies. However, the rules say the work cannot plagiarize other work, and must be original. Since the core of any AI requires this sort of tech, what is permitted?


Justin… my feeling is “no”, but let’s wait for the Entelect crew. Have you though about how would train it?

I would vote for “yes” for something like TensorFlow. I would just love to read a post after the competition on how Justin succeeded in training the net, since that is the bit I cannot wrap my head around so far.


I agree with gustav.
TensorFlow needs to be trained… which is not as easy as it seems.

So it could be classified as work to get something to deliver on a result that requires skill. I personally think it is a lot of effort to get a good result and might not be as much fun as programming. Though that is my opinion.

Also the changing ruleset and board-size will complicate the modelling.


Of course you may use TensorFlow, this is an AI Challenge afterall :wink:

As long as you do not resubmit someone else’s bot as if it is your own, and the bot exists because of your effort…then it is your own work

1 Like

What version(s) of Tensorflow will be supported?

I nominate 1.8 (pretty please)

Hopefully you can just upload a requirements.txt or something similar so that all the required libraries are pulled.

1 Like

This topic was automatically closed after 42 hours. New replies are no longer allowed.