We use cookies to ensure you have the best browsing experience on our website. Please read our cookie policy for more information about how we use cookies.
This challenge is really interesting, but the scoring system is broken. Assuming the signals are somewhat random, you can get a near 50% score by outputing the "correct" number of 1's or 0's. I think this problem would benefit from having say 20 test cases where, if your output reaches near 90% fidelity, that test case awards 10 points. Not only does this solve the nagging of never getting the 200pts, it stops potential cheating.
[Approximate Challenge] Convolutional Coding
You are viewing a single comment's thread. Return to all comments →
This challenge is really interesting, but the scoring system is broken. Assuming the signals are somewhat random, you can get a near 50% score by outputing the "correct" number of 1's or 0's. I think this problem would benefit from having say 20 test cases where, if your output reaches near 90% fidelity, that test case awards 10 points. Not only does this solve the nagging of never getting the 200pts, it stops potential cheating.