Paperback | $16.00 Short | £11.95 | ISBN: 9780262517348 | 120 pp. | 5.375 x 8 in | 19 illus.| January 2012
In Reliable Reasoning, Gilbert Harman and Sanjeev Kulkarni—a philosopher and an engineer—argue that philosophy and cognitive science can benefit from statistical learning theory (SLT), the theory that lies behind recent advances in machine learning. The philosophical problem of induction, for example, is in part about the reliability of inductive reasoning, where the reliability of a method is measured by its statistically expected percentage of errors—a central topic in SLT.
After discussing philosophical attempts to evade the problem of induction, Harman and Kulkarni provide an admirably clear account of the basic framework of SLT and its implications for inductive reasoning. They explain the Vapnik-Chervonenkis (VC) dimension of a set of hypotheses and distinguish two kinds of inductive reasoning, describing fundamental results about the power and limits of those methods in terms of the VC-dimension of the hypotheses being considered. The VC-dimension is found to be superior to a related measure proposed by Karl Popper, and shown not to correspond exactly to ordinary notions of simplicity. The authors discuss various topics in machine learning, including nearest-neighbor methods, neural networks, and support vector machines. Finally, they describe transductive reasoning and suggest possible new models of human reasoning suggested by developments in SLT.
About the Authors
Gilbert Harman is Stuart Professor of Philosophy at Princeton University and the author of Explaining Value and Other Essays in Moral Philosophy and Reasoning, Meaning, and Mind.
Sanjeev Kulkarni is Professor of Electrical Engineering and an associated faculty member of the Department of Philosophy at Princeton University with many publications in statistical learning theory.
"In their interesting and stimulating book Reliable Reasoning, Harman, a philosopher, and Kulkarni, an information scientist, illuminate the philosophical issues related to inductive reasoning by studying it in terms of the mathematics of probabilistic learning. One of the great virtues of this approach is that the inductive inference made through learning can survive changes in the probabilistic modeling assumptions. I find that the authors have made a convincing and persuasive case for rigorously studying the philosophical issues related to inductive inference using recent ideas from the science of artificial intelligence."
Sanjoy K. Mitter, Professor of Electrical Engineering, MIT
"This thoroughly enjoyable little book on learning theory reminds me of many classics in the field, such as Nilsson's *Learning Machines* or Minksy and Papert's *Perceptrons*: It is both a concise and timely tutorial 'projecting' the last decade of complex learning issues into simple and comprehensible forms and a vehicle for exciting new links among cognitive science, philosophy, and computational complexity."
Stephen J. Hanson, Department of Psychology, Rutgers University