To be fair, a machine learning algorithm often is built from generalized mathematical structures, but bear with me.
A lot of effort over the past decade has gone into image processing and pattern recognition. Particularly, many cutting edge solutions use networked models (not necessarily neural) with the intention of training an output node to be a reliable classifier. The generalized structure reflects a capacity to recognize patterns beyond images. I have often wondered if similar techniques could be applied to teach a computer high-level math such as formalized PM or category theory. After all, mathematics is often summarized as the study of patterns.
Can anyone point me towards some research that explores this sort of application? I think it would be fascinating.
[link][3 comments]
