• @NaiveBayesian
    link
    English
    44 months ago

    Whatever it takes to prove that √(log k/ε² + log log n) error bound with 95% probability – “probably almost correct” – what else do you need, really?