Machine Learning and Criminal Justice

I work with machine learning at a low level – I write neural network code and create machine learning algorithms. For people like me, there’s always a risk that I’ll lose site of the big picture. So, I go out of my way to seek out information about the uses of machine learning.

I came across information about an interesting company that creates software for use by people in the criminal justice system, in particular judges, district attorneys, and probation officers.

The company’s name is Northpointe. See The company’s Web site was pretty weak in the sense that the content was mostly marketing fluff. But by reading between the lines and finding some information on other Web sites, it was possible to figure out some of what Northpointe software does.


The example that I found most interesting is that judges, when sentencing convicted criminals, can use the software to get a probability estimate that the convicted person will commit more crimes (categorized by type of crime) after release.

Very interesting. Machine learning systems can be used to supplement human judgement in many areas. This is probably a good thing as long as the systems are just helpers and not making the final decisions. Humans are way better than computers at most tasks, including sentencing convicted criminals.

This entry was posted in Machine Learning. Bookmark the permalink.