Artificial Intelligence Researchers Produce “Data Science Machine” – Smarter Than Us

October 19, 2015 – In a machine versus human test a new artificial intelligence, deemed the “Data Science Machine” (DSM) outperformed 615 of 906 human teams. The Computer Science Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT) are the developers of this new tool which takes the humans out of data analysis. It features an algorithm entitled, Deep Feature Synthesis, which analyzes the relationships of data in a large database.

Determining what data to concentrate on and then finding patterns within a dataset is a difficult prospect for any machine let alone a human. Deep Feature Synthesis does both and does it faster than 2 out of every 3 human teams it competed against. The test involved enrolling DSM in three data science competitions focused on finding predictive patterns in unfamiliar large databases. DSM took a mere 2 to 12 hours to complete its analysis. Human teams took several months to accomplish the same analysis.

So how does DSM achieve its results? It is designed to understand database design – the storing of different information in tables with the use of numerical keys as an identifier to indicate correlation. DSM uses the correlation identifiers as a starting point in doing its analysis.

In two of the three contests, DSM’s predictions were 94 and 96% accurate. In the third DSM was 87% accurate. Max Kanter, one of the researchers on the project states, “we view the Data Science machine as a natural complement to human intelligence….there’s so much data out there to be analyzed. And right now it’s just sitting there not doing anything. So maybe we can come up with a solution that will at least get us started on it, at least get us moving.”

DSM, in analyzing datasets detects inference that can be derived from two or more different database sources. This makes it a useful tool for predicting buyer behavior patterns based on dissimilar information sources. It has the ability to correlate and test its assumptions recombining the data in different ways to optimize predictions.

States Margo Seltzer, Professor of Computer Science, Harvard University, “The Data Science Machine is one of those unbelievable projects where applying cutting-edge research to solve practical problems opens an entirely new way of looking at the problem….I think what they’ve done is going to become the standard quickly – very quickly.”