Dear participants,
This message is to notify you that, after consulting with the other members of the organizing committee, we came to the conclusion that algorithms that produce random solutions are not allowed: your algorithm may include some randomness for an initial state (in such case, provide the seed you used for generating your solution), but may not build the solution by blindly selecting pieces at random.
Kind regards,
San Vitale Challenge organizers
To further clarify our previous message, machine learning algorithms whose training process involves randomness (such as SVM, random forest, neural networks of any kind) are allowed.
Moreover, we allow algorithms that may have non-deterministic results and have some randomness involved, but some reasoning on the pieces' arrangement is present.
In both cases, please provide the seed and the pre-trained models if applicable, to ensure that the results we may generate are compatible with the ones you submit in the leaderboard.
On the other hand, algorithms that blindly select pieces to form a random solution (basically connecting pieces by guessing only) are not allowed, as we believe that these would not be useful in a real-use scenario.
If you have any questions please email us or post on this thread.