Chapter 11. Probability

Can we still expect anything, if chance is all there is?


Probability theory is one of the most beautiful subjects in mathematics, moving us back and forth between the stochastic and deterministic realms in what should be magic but turns out to be mathematics and its wonders. Probability provides a systematic way to quantify randomness, control uncertainty, and extend logic and reasoning to situations that are of paramount importance in AI: when information and knowledge include uncertainties, and/or when the agent navigates unpredictable or partially observed environments. In such settings, an agent calculates probabilities about the unobserved aspects of a certain environment, then makes decisions based on these probabilities.

Humans are uncomfortable with uncertainty, but are comfortable with approximations and expectations. They do not wake up knowing exactly how every moment of their day will play out, and they make decisions along the way. A probabilistic intelligent machine exists in a world of probabilities, as opposed to deterministic and fully predetermined truths and falsehoods.

Throughout this book, we have used probability terms and techniques as they came along and only when we needed them. Through this process, we now realize that we need to be well versed in joint probability distributions (for example, of features of data), conditioning, independence, Bayes’ Theorem, and Markov processes. We also realize that we can get back to the ...

Get Essential Math for AI now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.