JD and Harris on Artificial Intelligence

almondpitterpatterAI and Robotics

Feb 23, 2014 (3 years and 3 months ago)

66 views

Artificial Intelligence

And Machine learning

Drag picture to placeholder or click icon to
add

What is AI?

Different kinds of AI


Cybernetics and brain simulation


Symbolic


Sub
-
Symbolic


Statistical


Cybernetics and Brain Simulation


In the 1940’s and 50’s a number of researchers explored the
connection between Neurology information theory and
cybernetics.



some of the researchers used electronic networks to exhibit
rudimentary intelligence.


Some examples of this are W. Grey Walters “turtles” and
Johns Hopkins “Beast”

Symbolic


In the 1950’s when access to digital computers became
possible, AI research began to explore the possibility that
human intelligence could be reduced to symbol
representation.


Since the 1960’s, symbolic approaches had achieved great
success at simulating high
-
level thinking in small
demonstration programs.



Symbolic AI can be broken down into the following
categories: Cognitive simulation, Logic
-
based and
knowledge
-
based

Sub
-
Symbolic


In the 1980’s the progress in Symbolic AI had stalled and
many believed that symbolic systems would never be able to
imitate all the processes of human cognition.

Statistical


This type of research was started in the 1990s


It focused on advanced mathematical tools to solve specific
problems


The results of these tests were both measurable and variable
and have been some of the biggest contributors to many
success’s in AI research

Integrating the approaches


Intelligent agents


Multi
-

agent system

Tools to achieve this


Logic


Probabilistic methods for uncertain reasoning


Classifiers and statistical learning methods

Search and optimization


Many of the problems that AI face can in theory be solved by
searching through many possible solutions.


logical proof can be seen as searching for a path that leads
from premises to conclusions.


In the 1990s a different kind of search came to prominence, it
was based on the mathematical theory of optimization.

Logic


Logic is most commonly used for knowledge representation
and problem solving


There are a few different forms of logic used in AI research


Propositional logic, first
-
order logic, fuzzy logic, subjective
logic

Probabilistic methods for uncertain
reasoning


Many of the problems faced in AI research require an outside
source to operate with incomplete or uncertain information.



Some of the tools used to solve these problems use various
methods from probability theory and economics.


A key concept from the science of economics is "utility": a
measure of how valuable something is to an intelligent
agent, using this an agent can make choices and plan.

Classifiers and statistical learning
methods


The two simplest types of AI applications can be divided into
Classifiers and controllers.


An example of this is: Classifier (“ if round then ball”) and the
controller (“ if round then move”)


One of the most widely used classifiers is neural network

Evaluation of AI



Optimal: it is not possible to preform better.


Strong super
-
human: preforms better than all humans.


Super
-
human: preforms better than most humans


Sub
-
human: preforms worse than most humans

Drag picture to placeholder or click icon to add

The Future of Robotics and Artificial
Intelligence


https://www.youtube.com/watch?v=AY4ajbu_G3k

Drag picture to placeholder or click icon to add

Bibliograph


https://www.ai
-
class.com/

https://www.coursera.org/course/ml

http://www.academicearth.org/courses/ma
chine
-
learning