• If you were going to give a 2 hour survey talk on "Machine Learning" what topics would you cover?. Here's a brief but very interesting discussion on reddit. In the end, 2 hours aren't that much, so I would go with the "let's try to explain what it is and what it is not, what kind of problems can actually be solved and which ones are very probably a waste of time and money". May be even that will be too long.

  • The Field Guide to Data Science, a publication by Booz Allen Hamilton.

  • This article was making the rounds early this week: The age of ‘pre-crime’ has arrived. From the article: Los Angeles is considering sending “Dear John” letters to the homes of men who solicit prostitutes hoping the mail will be opened by mothers, girlfriends or wives. Basically, what the city council is planning to do is to scan license plates from drivers that venture themselves into neighborhoods where prostitution typically takes place, and then using the registered data to send the letters. This is not really pre-crime, but rather a certainly controversial use of citizen data.

There is no comment system. If you want to tell me something about this article, you can do so via e-mail or Mastodon.