The Cerebellar Articulated
Artificial Controller (CMAC):
Neural Architecture: Theory and Software Implementation
Department of Mathematical Sciences Clemson University

James K. Peterson

1995:
Revised 1998:
Revised 1999:
Revised 2001:
Generated 2/8/2001

• Introductory Remarks:
  • Copyright Stuff:
  • FTP Access:
  • How to Build the Code:
  • Code Snippets or How to Whet Your Appetite:
  • The Genesis of This Code:
  • Candid Assessments of The Code:
  • Future Work:
  • Acknowledgements:
• Neurobiological Underpinnings:
  • The Purkinje Cell Wiring Diagram:
  • The Purkinje Cell Abstraction:
• A CMAC Primer:
  • The Raw Input Space:
    • The Zeroth Level:
  • Adding the Next Level:
  • The General Model: L Levels:
  • A Specific Example:
  • Function Approximation Using the CMAC:
  • Hashing Strategies:
  • The Purkinje Model:
• CMAC Architectures:
  • The CMAC Architecture:
  • Training the CMAC Architecture:
  • Comments on Training:
  • Working Memory Via Hashing:
  • Implementing Scalar CMAC Architectures:
  • Implementing Vector CMAC Architectures:
  • Future Extensions:
• CMAC Convergence:
  • Introduction:
  • CMAC Training:
  • The Matrix Form:
    • The CMAC Training Generator Matrix:
    • The Structure of D:
    • The Structure of C_i:
  • Convergence Analysis:
    • The Structure of V^T:
  • Quality of Solution:
  • Solution Magnitude and Range Estimates:
  • Conclusion:
• A Guide to the Software CMAC Implementation:
  • Basic Setup:
  • Some Sample Code:
    • Using the Explicit Constructor:
    • Using The Default Constructor:
• The Structure of the CMAC Class:
  • The Makefile: MakeLibrary:
  • The CMAC Class:
    • The CMAC Class header:
  • Construction of the CMAC Object:
    • Explicit Constructor: constructor.c:
    • Default Constructor: default_constructor.c:
    • Copy Constructor: copy_constructor.c:
    • Configuration Function: config.c:
    • Get Configuration Files: cmacfile.c:
    • Internal Allocation: getcmac.c:
    • Initialization: cmacinit.c:
  • Destruction of the CMAC Object:
    • Destructor: cmacfree.c:
  • Evaluation Functions for the CMAC Object:
    • Evaluation: cmaceval.c:
    • Hashed Memory Computations: virtmem.c:
    • CMAC Training Algorithm: train.c:
  • Utility Functions for the CMAC Object:
    • Size Calculations: number.c:
    • Echo CMAC Class Parameters: echo.c:
    • Hash Algorithm: func.c:
    • Compute RMS Error: getrms.c:
    • getworkingaddress.c:
    • getlevels.c:
    • getnumberintervals.c:
    • getinputsize.c:
    • getoutputsize.c:
    • getminx.c:
    • getmaxx.c:
    • getwidth.c:
    • loadinputsize.c:
    • loadoutputsize.c:
    • loadtrainsize.c:
    • loadtestsize.c:
    • gettestsize.c:
    • gettrainsize.c:
    • Get Test Data: readte.c:
    • Get Training Data: readtr.c:
    • Get Stored Parameters: readw.c:
    • Save CMAC Parameters: writew.c:
• Simulation Runs: Function Approximation:
  • The Application Code:
    • The MakeFile: MakeApplication
      • The Make File for test_generator.c
      • The Make File for simul.c
      • The Make File for testmodel.c
    • The Application Code:
      • test_generator.c
      • simul.c
      • testmodel.c
    • Configuration Files:
  • 64 Level Approximation:
  • 20 Level Approximation:
  • Eight or Fewer Level Approximation:
• References:

This document was tranlated from L^AT_EX by H^EV^EA and H^AC^HA.