Turing’s Connectionism, 2002 An Investigation of Neural Network Architectures Discrete Mathematics and Theoretical Computer Science Series

Alan Mathison Turing (1912-1954) was the first to carry out substantial re­ search in the field now known as Artificial Intelligence (AI). He was thinking about machine intelligence at least as early as 1941 and during the war cir­ culated a typewritten paper on machine intelligence among his colleagues at the Government Code and Cypher School (GC & CS), Bletchley Park. Now lost, this was undoubtedly the earliest paper in the field of AI. It probably concerned machine learning and heuristic problem-solving; both were topics that Turing discussed extensively during the war years at GC & CS, as was mechanical chess [121]. In 1945, the war in Europe over, Turing was recruited by the National Physical Laboratory (NPL)! in London, his brief to design and develop an electronic stored-program digital computer-a concrete form of the universal Turing machine of 1936 [185]. Turing's technical report "Proposed Electronic 2 Calculator" , dating from the end of 1945 and containing his design for the Automatic Computing Engine (ACE), was the first relatively complete spec­ ification of an electronic stored-program digital computer [193,197]. (The document "First Draft of a Report on the EDVAC", produced by John von Neumann and the Moore School group at the University of Pennsylvania in May 1945, contained little engineering detail, in particular concerning elec­ tronic hardware [202].

1. Introduction.- 1.1 Turing’s Anticipation of Connectionism.- 1.2 Alan Mathison Turing.- 1.3 Connectionism and Artificial Neural Networks.- 1.4 Historical Context and Related Work.- 1.5 Organization of the Book.- 1.6 Book Web-Site.- 2. Intelligent Machinery.- 2.1 Machines.- 2.2 Turing’s Unorganized Machines.- 2.3 Formalization and Analysis of Unorganized Machines.- 2.4 New Unorganized Machines.- 2.5 Simulation of TBI-type Machines with MATLAB.- 3. Synthesis of Logical Functions and Digital Systems with Turing Networks.- 3.1 Combinational versus Sequential Systems.- 3.2 Synthesis of Logical Functions with A-type Networks.- 3.3 Synthesis of Logical Functions with TB-type Networks.- 3.4 Multiplexer and Demultiplexer.- 3.5 Delay-Unit.- 3.6 Shift-Register.- 3.7 How to Design Complex Systems.- 3.8 Hardware Implementation.- 4. Organizing Unorganized Machines.- 4.1 Evolutionary Algorithms.- 4.2 Evolutionary Artificial Neural Networks.- 4.3 Example: Evolve Networks that Regenerate Bitstreams.- 4.4 Signal Processing in Turing Networks.- 4.5 Pattern Classification.- 4.6 Examples: Pattern Classification with Genetic Algorithms.- 4.7 A Learning Algorithm for Turing Networks.- 5. Network Properties and Characteristics.- 5.1 General Properties.- 5.2 Computational Power.- 5.3 State Machines.- 5.4 Threshold Logic.- 5.5 Dynamical Systems and the State-Space Model.- 5.6 Random Boolean Networks.- 5.7 Attractors.- 5.8 Network Stability and Activity.- 5.9 Chaos, Bifurcation, and Self-Organized Criticality.- 5.10 Topological Evolution and Self-Organization.- 5.11 Hypercomputation: Computing Beyond the Turing Limit with Turing’s Neural Networks?.- 6. Epilogue.- Useful Web-Sites.- List of Figures.- List of Tables.- List of Examples, Theorems, Definitions, Propositions, and Corollaries.

Contains a Foreword by B. Jack Copeland and D. Proudfoot Develops Turing's ideas, by building his machines and training them using evolutionary algorithms Goes beyond Turing's ideas, and proposes new machines