What I've Been Reading

The greatest benefit of machine learning may ultimately be be not what the machine learns but what we can learn by teaching them.

The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World by Pedro Domingos.

A must read for students, decision makers and public in general. Although Professor Domingos explains the algorithms as interpretable as possible, it's not an easy read especially the middle chapters. Having said that this is the only popular book available for public to enlighten themselves on machine learning. I have personally struggled for the past couple of years learning machine learning via hard math based books or small but brilliant blog posts. I wish Professor Domingos wrote this book earlier.

And finally, according to him; this is one plausible version of the Master Algorithm:

Let's use the name Alchemy to refer to our candidate universal leaner for simplicity. Alchemy addresses Hume's original question by having another input besides the data: your initial knowledge, in the form of set of logical  formulas, with or without weights. The formula can be inconsistent, incomplete, or even just plain wrong; the learning and probabilistic reasoning will take care of that. The key point is that Alchemy doesn't have to learn from scratch. In fact, we can even tell Alchemy to keep the formulas unchanged and learn only the weights. In this case, giving Alchemy the appropriate formulas can turn it into a Boltzmann machine, a Bayesian network, an instance-based learner, and many other models. This explains why we can have a universal learner despite the "no free lunch" theorem. Rather, Alchemy is like an inductive Turing machine, which we can program to behave as a very powerful or a very restricted learner; it's up to us. Alchemy provides a unifier for machine learning in the same way that the internet provides one for computer networks, the relational model for databases, or the graphical user interface for everyday applications.