June 11, 2018
As of 2016, a rough rule of thumb is that a supervised deep learning algorithm will generally achieve acceptable performance with around 5k labeled examples / category and will match or exceed human performance when trained with a dataset containing at least 10M labeled examples. https://t.co/B7TELpx707
Original tweet: https://twitter.com/giano/status/1006199100977680384
This is one of the many thoughts I post on Twitter on daily basis. They span many disciplines, including art, artificial intelligence, automation, behavioral economics, cloud computing, cognitive psychology, enterprise management, finance, leadership, marketing, neuroscience, startups, and venture capital.
I archive all my tweets here.
I archive all my tweets here.