Skip to content

Comment on Intel introduces 3-GHz desktop chip (2002)parent

Comments

Unfortunately, many machine learning algorithms aren't easily parallelizable. Consider any boosting ensemble learner, such as the popular random forest, which optimizes the result from the previous iteration. (I run the machine learning startup diffbot.com and we could really use faster cycles.)

what sort of algorithms would you like to have faster variants/implementations of?

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.