The Curse of Dimensionality

The curse of dimensionality is a statistical principle that relates back to basically having that many more dimensions in your data. Now you might think that higher spatial resolution is always a better thing, but it has been shown in the past that you sometimes get better performance out of your classifier if you have maybe lower resolution data. There's just something about when you squint at something and suddenly a pattern comes out of it. It's similar when you're making the computer squint at the data and somehow that teases out the information you're looking for.