More than 135,000 people around the world have signed up for Stanford University’s upcoming Introduction to Artificial Intelligence online class. The course will be taught by two leading thinkers in the AI field, Sebastian Thrun, a professor at Stanford, and Peter Norvig, director of research at Google. Clearly, this shows the allure of Artificial Intelligence or free education, or both!
Additionally, Introduction in Machine Learning is also available
I came across an interesting WSJ article on an experimental service from Google called “Google Correlate“. In a nutshell, you can use the tool to see if arbitrary items correlate well against the other. One of the examples demonstrated (with a dose of tongue-in-cheek humor) is a correlation between the Federal Reserve’s balance sheet and the number of Google searches on “nausea remedies”. They do indeed correlate well, until you realize that the correlation to the Google search “how to get over a guy” actually correlates better. While there certainly is power in correlating data sets against each other, one can obviously see that relationships in data could be more coincidental than truly being related.
There is an obvious analogy here with APM (Application Performance Management) data. Humans often have to “eyeball” charts of time-series data, looking for things that “correlate”. And even if you could use basic correlation mathematics to compare data, obviously you can still get “coincidental” correlations.
That is why more intelligent analytics are needed to diagnose application problems. Prelert’s 3rd-generation self-learning it analytics uses a unique blending of multiple, modern algorithms that have proven, time in again, to find the things that are truly related in the data, not those that are merely coincidental
