Logic for Learning: Learning Comprehensible Theories from Structured Data

Citation

Lloyd, J.W. (2004), "Logic for Learning: Learning Comprehensible Theories from Structured Data", Assembly Automation, Vol. 24 No. 3, pp. 325-325. https://doi.org/10.1108/aa.2004.24.3.325.3

This book forms part of the Springer series of Cognitive Technologies, which addresses subjects including natural‐language processing, high‐level computer vision, cognitive robotics, automated reasoning and knowledge representation. Logic for Learning combines high‐order computational logic with machine learning. It is suitable for students and researchers alike and does not require previous knowledge of either main subject.

Chapter 1 provides an Introduction to logic and learning, and an outline of the book. Chapter 2 discusses Logic, including types, type substitutions, terms, λ‐conversion, and model and proof theory. Default and normal terms, an equivalence relation on normal terms, a total order on normal terms, and metrics and kernels on basic terms, are addressed in Chapter 3, Individuals.

Predicates and Computation are presented in Chapters 4 and 5, respectively. Subjects discussed include: transformations, standard and regular predicates, efficient construction of predicates, program as equational theories, and programming with abstractions. The final chapter of the book, Learning, addresses the problem of learning comprehensible theories from structured data.

Learning for Logic is clearly presented, but may prove heavy reading for those who are less than enthusiastic about computational logic or machine learning.