Over the long weekend, I started reading a...

Over the long weekend, I started reading a new book, Human Compatible: reasonably interesting so far! Discussions about artificial intelligence are always intriguing, but it's especially interesting to hear about it from a perspective of years of practice that can generalize what's happening well. " "" "It also made me thankful for the coursework I've done in AI. The discussion of Bayesian networks reminded me of several classes I've taken during my academic career and that the inventor of Bayesian networks taught at my alma mater, UCLA. " "" "If you've never heard of them before, Bayesian networks are probabilistic models atop directed, acyclic graphs (DAGs). Here, you can see an example of a DAG, representing the model to help explain your current symptoms: are they because of the common cold or the martian death flu?" "" "Through Bayesian inference, it is possible to evaluate potential sources of symptoms based on prior known probabilities. While Bayesian networks may be too complicated for some, the underlying probabilistic reasoning is something we all already do; Judea Pearl, a professor at UCLA, was able to formalize it into mathematical notation. We reason all the time probabilistically as humans, so learning more about how that kind of reasoning can work can't hurt! #bendevtip

---

Original post on LinkedIn