USER'S MANUAL FOR THE CALM LEARNING-MACHINE SIMULATION PROGRAM.
Abstract
The CALM (Collected Algorithms for Learning Machines) program system is a means of implementing a wide class of learning machines on a digital computer. A learning machine is a trainable pattern classifier, built around an array of adaptive logic elements, that categorizes patterns on the basis of its past experience. The CALM system provides a framework for learning-machine simulation largely independent of the computer used and the types of patterns being classified. The actions performed by CALM include training a learning machine on a set of patterns, testing the classification error rate of the trained machine on the same or an independent set of patterns, and modifying, saving, and restoring the machine's memory and control parameters. The report is a user's manual for CALM, and specifically for the CALM system implementation for the General Electric 635 computer. The report covers the nature of the learning machines realized by CALM, the CALM program itself, and the use of CALM in learning-machine experimentation. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1967
- Accession Number
- AD0648959
Entities
People
- John H. Munson
Organizations
- SRI International