Probably Approximately Correct: Nature's Algorithms for Learning and Prospering in a Complex World (Paperback)
“If you want to succeed
in a job interview…or in choosing a life partner, you can be quite sure that
there is no equation that will guarantee you success.” What we actually
utilize in these everyday situations that are not governable by explicit mathematically
based theories are algorithms that “run in environments unknown to the designer
and…learn by interacting with the environment how to effectively act in
it.” He makes a valiant go
at defining and quantifyingthese
pragmatic, “common sense” strategies, which include generalizing as well as
reasoning and learning. The
model of reasoning presented is simple but powerful: we manipulate (using his
concept of “robust logic”) a limited number of concepts within our working
memory, all of which come from either the complex outside world or our
long-term memory. He does acknowledge the main obstacle to this work: “…the
learning algorithms that are hard-wired in the brain are yet to be
learned.” The application
of his theories to biochemistry and behavior in an evolutionary context seem
premature, specifically because how the former gives rise to the latter is not
understood at present and may not be tractable. He gives a nice overview of the
limits of computation and machine learning while not employing very much math
or technical jargon. The
philosophical extrapolations near the end of the book are quite good as well.
How does life prosper in a complex and erratic world? While we know that nature follows patternssuch as the law of gravityour everyday lives are beyond what known science can predict. We nevertheless muddle through even in the absence of theories of how to act. But how do we do it?
In Probably Approximately Correct, computer scientist Leslie Valiant presents a masterful synthesis of learning and evolution to show how both individually and collectively we not only survive, but prosper in a world as complex as our own. The key is probably approximately correct” algorithms, a concept Valiant developed to explain how effective behavior can be learned. The model shows that pragmatically coping with a problem can provide a satisfactory solution in the absence of any theory of the problem. After all, finding a mate does not require a theory of mating. Valiant's theory reveals the shared computational nature of evolution and learning, and sheds light on perennial questions such as nature versus nurture and the limits of artificial intelligence.
Offering a powerful and elegant model that encompasses life's complexity, Probably Approximately Correct has profound implications for how we think about behavior, cognition, biological evolution, and the possibilities and limits of human and machine intelligence.