Please enable JavaScript.
Coggle requires JavaScript to display documents.
Artificial Intelligence: Modeling the Brain (Semantic Information…
Artificial Intelligence: Modeling the Brain
Intelligent Machines
Early Programs
Human Logic
Logic Theorist Program
The General Problem Solver (GPS)
limited domains
Expert Systems
Applications
PUFF
Medical system
for diagnosis of
respiratory conditions
PROSPECTOR
Used by geologists to identify sites for drilling or mining
MYCIN
Medical system for
diagnosing blood
disorders
DESIGN ADVISOR
Gives advice to
designers of chips
processor
DENDRAL
Used to
identify the structure of
chemical compounds (1965)
LITHIAN
Gives advice
to archaeologists examining stone tools
Natural Language Processing
translate
ordinary human naturallanguage into language
computers
complexity of human
language
Attempt at NLP: ELIZA (1965)
using a person's
responses to shape the computer's
replies
Javascript Version of ELIZA
If key words have not been stored, ELIZA cannot process it!
Lack ability to Learn, Reason,Plan, Understand
Semantic Information Processing
Terry Winograd
1972
programme
which combines grammar, semantics, and
reasoning in an intimate way
SHRDLU
Created by Terry Winograd in 1972
A simple virtual/simulated robot
Has natural language processing capabilities
based on the expert system idea
A hand & An eye
Ability to manipulate toy blocks on a table
Carries out command given by human
SHRDLU contains:
A dictionary
syntax
semantics
grammar
A parser
Set of programmes to control the responses made by SHRDLU
Script Applier Mechanism (SAM)
designed to
understand storiesthat rely heavily on
scripts
Story “Understander
Makes inferences about stories
Artificial Intelligence: Linguistics & Artificial Intelligence
Preliminary Insights
Programmed to perform human task
Machine Translation
machine perform simple linguistic task (1960)
Automatic translation
Input (Source language)
Output (Target language)
Memory (Words of Source Language & Their Equivalent in Target
Language)
Difficulty with translation
meaning
Machine translation works best only when vocabulary involved
limited to very specific domain
Natural Language Processing
sentence pattern frames stored as data
Translate ordinary human commands into language
computers can understand & act on
The goal is “to achieve human-like language processing”
Renown program = ELIZA
Matching input to fixed pattern
collect data
parse
repository
view results
optimize
apply text mining algorithms
1 more item...
Machine Learning :Connectionism & the Neural Network
Biological Neural Networks
The Nerves System: Neurons
Neurons = nerve cells, specialized to
receive and
transmit
information in the nervous system
Receive inputs and carry
them to the cell soma
Main body of the cell
(i.e. cell soma) & contains the nucleus
(generic material of neuron – DNA)
Artificial Neural Network (ANN)
Consists of nodes-input-output
idea?
Links (connection between
nodes)
On
set to fire at a particular time
will send a signal
to its neighbours,via the links
some links will conduct & some won't
off
receiving a signal will turn on & the
sending node, having fired, will turn off,
Weights (numeric value)
Network (for activation)
Neural networks
connections to produce
dynamic memory
incorporates the above-stated learning
properties
Learning by repetition
More repetition --> more conductance
Less use --> less conductance
Unpredictability
probability of firing
Associative learning
A node that fires causes the firing of another, which connects
with the conducting link
Forgetting from disuse
Early developments in neural net research
Donald Hebb’s
basis of artificial neural nets
Perceptrons
Is a hypothetical nervous system realised as a neural net
Stored information represents experience
is a neural net that learns from
experience
simple form of supervised learning
Robotics
Machines to perform specific
functions
Essential Features
Sensing (Sensors)
Robot Vision (eyes)
Face Detection in
Color Images
Fingerprint
Matcher
touch and pressure sensors
(hands)
chemical sensors (nose)
hearing and sonar sensors
(ears)
taste sensors (tongue)
Movement
rolling on wheels
walking on legs
propelling by thrusters
Energy (Power)
solar powered
electrically powered
battery powered
Intelligence
Programming = so that it
knows what it is to do
Automated Speech Recognition
Artificial life: Animats
To exhibit at least some of the
survival capacities of real animals
Based on behaviour
the (mechanical) form of an
insect
place it in a real-world
environment
endow it with sensors to detect the
environment
Build a neural net into it
Cog Project
Building a Humanoid Robot = resembling Human