Download COSC 480: Genetic Algorithms in Machine Learning

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Y chromosome wikipedia , lookup

X-inactivation wikipedia , lookup

Koinophilia wikipedia , lookup

Genome (book) wikipedia , lookup

Neocentromere wikipedia , lookup

Chromosome wikipedia , lookup

Gene expression programming wikipedia , lookup

Population genetics wikipedia , lookup

Polyploid wikipedia , lookup

Karyotype wikipedia , lookup

Microevolution wikipedia , lookup

Transcript
George Yauneridge

Machine learning basics

Types of learning algorithms

Genetic algorithm basics

Applications and the future of genetic
algorithms

Machine learning is a topic of artificial
intelligence

The focus is on developing and implementing
algorithms that allow machines to “learn”

For machines, learning is acquiring new data
and forming decisions based on all available
data

Case Based Reasoning
 Modifies the solution from a past problem

Decision Trees
 Analyze a tree of conditions to give yes/no answers

Data Mining and Pattern Recognition
 Finds patterns in large amounts of data

Neural Networks
 Made of many units, each capable of input/output

Reinforcement Learning
 The system analyzes its interactions with its
environment, usually involving trial and error

Inductive Logic
 Imitates human interpretation of data

Use DNA and evolution from biology as a
model

Use a population of parent solutions that
compete to produce children

Only the strongest solutions pass on their
information

A each solution is called a chromosome

Carries the information

Several ways to encode the data
 Examples
C1: 011001110101
C2: 011010110100

“Strong” chromosomes are selected to pass
on their information

Strength of a chromosome is determined by
a fitness function

Imitates Darwin’s survival of the fittest theory

Chromosomes switch parts of their
information to form children

Several different methods
C1: 011001110101
C2: 011010110100
Crossover
C1: 01100111o100
C2: 011010110101

Certain bits are inverted

Imitates mutations that would occur in nature
C1: 011001110101
Mutation
C1: 010001110111



Create a random population of chromosomes
Evaluate the fitness of each chromosome
Create the next population






Select 2 fit parents
Perform crossover
Perform mutation
Place the children in the new population
Using the new population, test if the end
condition is met
Loop
http://www.obitko.com/tutorials/genetic-algorithms/example-functionminimum.php

Evolving music

Strategy planning

Evolving programs

Other evolutionary and decision based
applications
Association for the Advancement of Artificial
Intelligence. http://aaai.org
AI Horizon. http://aihorizon.com
Marek Obitko.
http://www.obitko.com/tutorials/geneticalgorithms
CMU CS Department.
http://www.cs.cmu.edu/afs/cs/project/airepository/ai/html/faqs/ai/genetic