DMLS a machine learning system.

number: 
240
English
department: 
Degree: 
Imprint: 
Computer Science
Author: 
Mohammed Abbas Fadhel
Supervisor: 
Dr. Riyadh A.Mehdi
year: 
1996
Abstract:

The Principal objective of this research is to introduce a machine learning system named Developed Machine Learning System (DMLS) which uses learning from examples method. The designed system adopts the version space model. However, the version space model suffers from two shortcomings : Noisy training instances. Disjunctive concepts. In this work, a solution to overcome the second problem was suggested and implemented successfully. DMLS uses a predicate logic form of language to interact with the user of this system. DMLS chose this form of language interface because the predicate logic form is easy to understand, simple. DMLS consists of four main modules: learning, dialog, manage-ment and translation. DMLS can be successfully used in building expert systems that have no data base at the beginning of their operation, and depends on automatic knowledge acquisition to acquire their knowledge. DMLS was implemented on an IBM/PC compatible using Turbo Prolog 2.0.