Eric Mittelhammer
About Contact

Mar 15, 2015

Handwriting Recognition: Neural Networks vs. Distance Metrics

If you’ve ever taken a course on machine learning or artificial intelligence, the unit on neural networks most certainly had an exercise on handwriting recognition. There’s a reason for this: they’re very good at generalizing problems that have extremely fuzzy input.

  • machine learning
  • neural networks

Feb 11, 2015

What's the Deal with Amazon Kinesis?

The Amazonian marketing angle on Kinesis is pretty clear. The product page is a treasure trove of buzzwords: “Streaming”, “Real-Time”, “Parallel”, “Magic” (ok i threw that one in there myself). Great for CTOs, not so much for developers. The documentation is a bit sparse as well. All the good stuff is buried pretty deep and is spread out amongst the developer guide, api reference, and client library documentation. Hopefully this article will help you grok it all a little better.

  • AWS
  • Kinesis
  • big data

Jun 14, 2014

Clustering with TF-IDF

Originally published at: http://tech.kinja.com/the-kinja-family-tree-1590442849

  • machine learning
  • NLP
  • clustering
  • R

Feb 21, 2014

Actors don't play nice with value classes

Scala value classes are a sugary-sweet convenience, but they can introduce some low-level conflicts due to their compilation magic.

  • scala
  • akka
  • actors
Newer
Older