News for hackers. Stuff that matters.
Re:Ask HN: Why do so many startups claim machine learning is... by killjoywashere on Wednesday November 13, 2019 @05:27PM (#21530375)
Um, yes and no. Some of the data annotation problems get really hard and really expensive. Look at radiology or genomics. You literally need N people to die for more knowledge in some of these situations.
Re:Ask HN: Why do so many startups claim machine learning is... by jnwatson on Wednesday November 13, 2019 @05:49PM (#21530579)
Data cleaning and data labeling are two separate tasks, with vastly different workflows. For example, getting structured data out of Whois requires a lot of cleanup. Labeling domains as “bad” might take forever.