You are on page 1of 8

INTRODUCTION TO

ARTIFICIAL INTELLIGENCE
FOR IT & NON-IT PROFESSIONALS
SOLVING REAL WORLD PROBLEMS BY LEARNING
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• Algorithms have been developed that are effective for certain types of
learning tasks, and many significant commercial applications have
begun to appear
• In speech recognition, the algorithms based on Machine Learning
outperform all other approaches that have been attempted to date
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• Applications that use learning algorithms include:
• Spoken digits and word recognition
• Handwriting recognition
• Path finders
• Intelligent homes
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• Any given learning problem is primarily composed of three things:
1) Input
2) Processing Unit
3) Output
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• The Input is composed of examples that can help the learner learn the
underlying problem concept
• Example: to build a learner for recognizing spoken digits
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• The Processing Unit is the learning agent in our focus of study
• Any learning agent or algorithm should, in turn, have at least the
following three characteristics:
1) Feature Representation
2) Distance Measure
3) Generalization
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• Feature Representation:
• The input is usually broken down into several features
• Example: the difference between a good and not so good student
• Distance measure
• Given two different inputs, the learner should be able to tell them apart, or
differentiate
• The distance measure is the procedure that the learner uses to calculate the
(various) difference between the two inputs.
SOLVING REAL WORLD PROBLEMS BY
LEARNING
• Generalization
• In the training phase, the learner is presented with some positive
and negative examples from which it learns
• In the testing phase, when the learner comes across new but
similar inputs, it should be able to classify them similarly. This is
called generalization

You might also like