How UCLA PhD candidate used machine learning to find true love

By Valentino in Algorithms, Machine Learning, Statistics

 

Just before Chris Mckinley earned his PhD in mathematics from UCLA, he was on a quest for love.  Like millions of others on that quest, he turned to the aid of dating sites.  After countless failed matches and little to no prospects he turned to his skills to help him do the heavy lifting.  Mickinley’s research as a PhD candidate focused on large-scale data processing and parallel numerical methods.  Knowing that one of the dating sites became famous for their mathematical approach to matching couples, he created a few python scripts to data mine potential matches that were more suited to his him that what the site was matching him with.  Using a clustering algorithm based on one developed by Bell Labs (previously used to analyze diseased soybean crops) called K-modes, he was able to reverse-engineer how the dating site weighted the questions it asks each person to gauge how well they match with another.  With this data in hand, McKinley then used a machine-learning called adaptive boosting to weight his own answers to these questions to better match himself with suitors that more accurately fit who he’d like to meet.  Needless to say, his mathematics brought him a fairytale ending.

Found on: Wired.

The amazing accuracy of GE’s new Revolution CT scanner

By Valentino in Innovation

Full Story from GE

Stop junking up your .hosts file

By Valentino in Code

*.laclhost.com, there's no place like 127.0.0.1

I hate having to constantly manage my .hosts file. Even with the PrefPane helpers, POW, an the countless other tools for linking your development application to your browser. Thankfully, there’s lacolhost. I can finally sleep at night without dreading starting up a new project.

All it takes is a nice visit to lacolhost.com, and it will redirect right back to my localhost. Now I can test my app with subdomains without having to create the link for it every time. Awesomeness.