Send your feedback or ask for help here!
Very educational puzzle, clear description, nice youtube links - so congrats, I liked this!!!
The puzzle itself is not ‘very hard’ at all, we just need to follow the instructions in the statement carefully. Not getting lost with such high number of different variable names and array indexes was the main challenge. The ability to play around with the modell in puzzle part-2 was also nice. I found some good parameters after 5 minutes of try-and-error.
Interestingly, having a deeper NN or with more nodes per layer did not always improve the results, so in the end, modifying Eta and maxing out the # of training iterations within the timeout threshold proved the way to go for me.
Is there some rules of thumb to choose the correct number of hidden layers, nodes per layers or eta? Are more training runs always better or some overfitting for noise can kick in?
If you are interested in neural networks or machine learning in general, I recommend this free course:
The backpropagation algorithm (topic of this CG puzzle) is discussed in detail on Week 4 & 5.
The course involves lecture videos, handouts and programming assignments in Octave. (Octave is similar to Matlab, but free, for the assignments usually just a few lines of code are needed, the necessary basic Octave skills can be learned in no time. If the course were newer, probably they would have used Python + numpy, but that is a nuissance - the focus is on understanding how the most important ML algorithms work and not how to code.)
Hi, I am surprised that this is the only discussion thread I find about the two puzzles Binary Neural Network parts 1 & 2. There is no “discussion” tab in either puzzle…
I would just like to say thank you to player_one because this was a great way to learn about neural network. The first puzzle is well guided and can be solved just by following the instructions step by step. And for the second, I expected huge headaches and in fact, copy/pasting the code of the first puzzle and just a little bit of tuning and tries have solved it !
This was a great experience and shows to a total newbie the potential of this technique.
Thanks a lot