Hello, folks.

The concepts of evolution and adaptation
are ubiquitous in complex systems.

As we'll see in upcoming units, the
mechanisms of Darwinian natural selection

are relevant not only
for biological organisms,

but turn out to be central for social and
technological systems as well.

Computer scientists have been interested
in simulating biology

for as long as computers
have been around.

In this unit, I'll cover one approach
to biologically-inspired computation,

called genetic algorithms.

These algorithms were initially developed
as a way of modelling evolution and

adaptation across disciplines,
but they've also turned out to be

very successful as
artificial intelligence methods,

a way of evolving programs and other
computational objects,

without having to engineer them ourselves.

Sometimes, the results of these algorithms
are very surprising.

They come up with things that no human
would have thought of.

And sometimes the solutions
that they come up with

are more successful than those
designed by humans.

In this unit, I'll cover how
genetic algorithms work,

and how they've been used
in the real world.

Let me start by outlining the basics
of evolution by natural selection.

First, organisms inherit traits
from their parents.

The traits are inherited 
with some variation,

via mutation, sexual recombination,
and other variation-generating mechanisms.

But due to competition for limited
resources, the organisms

best adapted to the environment tend
to produce the most offspring.

This was Darwin's insight into the idea
of natural selection.

Selection is due to competition.

In that way, traits producing adaptable
individuals spread in the population.

Early on,
computer scientists inspired by biology,

saw that this could all
happen not just in nature,

but also inside computers, 
where computer organisms,

such as programs, could play the role of
biological organisms,

and could evolve via
Darwinian evolution.

One of the first people 
to develop this idea in any detail,

was a computer scientist,
John Holland,

at the University of Michigan,
who in the 1960's and the 1970's,

developed the idea of genetic algorithms,
also known as GA's.

Holland introduced the notion of GA's
in his book,

"Adaptation in Natural and 
Artificial Systems", published in 1975.

This book presented a mathematical 
framework for understanding

adaptation in a very general way,
one that went beyond

biological adaptation and 
encompassed social, technological, and

artificial adaptation in computers.

The last part lead to the development of
genetic algorithms and their subsequent

application in many real-world problems.

Here's some examples of 
real-world use of genetic algorithms.

This is just to give you a sense of 
of the diversity of applications.

GA's have been used to design 
parts of aircraft by GE and Boeing.

Also by NASA to design antennae for 
spacecraft.

John Deere and Company have used
genetic algorithms to optimize

assembly line scheduling.

They've been used by several companies
to assist in automated drug discovery.

GA's have been used to learn
to detect fraud in

financial transactions and
financial trading.

They've been used to develop programs
for the automated analysis

of satellite images at Los Alamos,

and they've been used to generate

realistic computer animation
in several movies,

including
"Lord of the Rings" and "Troy".