Professional Documents
Culture Documents
§ What is AI ?
§ How it evolved ?
§ What can we do with it ?
§ How can we use AI ?
A Chinese Proverb
Ability to…
§ Think
§ Analyse
§ Learn
§ Reason
§ Do creative work
§ Perception
§…
What is Intelligence ?
INTERNAL
INPUTS PROCESSES
Senses environment Has knowledge
Has understanding/
intentionality
See
Can Reason
Hear
Touch
Taste
Smell Exhibits behaviour
Mind
OUTPUTS
What is AI ?
§ Media creates the world today
§ AI =
§ 2001: Space odessey
§ Star Trek
§ AI: The movie
§ Often portrayed as
§ A property of an evil computer
§ Computers doing impossible things
§ Books and movies
§ Inspired many AI researchers
§ Raised the expectations of general public
What is AI ?
What others have to say
What is AI ?
§ Not necessarily
§ Sometimes more…sometimes less
§ It is like
§ Physical Vs Electronic Books
§ Actual Vs Virtual shopping
§ Bird Vs Airplanes
Scientific Goal:
To understand computational
mechanisms needed for
modeling intelligent behaviour
Goals of AI
Intelligent Systems
Behaviourist View on Intelligent Machines
§ Many scientists believe that only things that can be directly observed
are “scientific”
§ This view can be summarized as:“If it walks like a duck, quacks like a
duck and looks like a duck - it is a duck”
A brief history of AI
§ Gestation (43-56):
§ automata theory, neural networks, checkers, theorem proving.
§ Shannon, Turing, Von Neumann, Newell and Simon, Minsky,
McCarthy, Darmouth Workshop.
§ Great expectations (52-69):
§ computers can do more than arithmetics!
§ General Problem Solver (GPS), better checkers
§ LISP (LISt Processing language)
A brief history of AI
A brief history of AI
§ Behaviour
§ Chinese questions go into the room.
§ Chinese answers come out of the room.
§ Functionalist interpretation
§ Whoever is in the room
understands Chinese.
Same situation with our Students ?
Do we really understand … or
Just CUT & PASTE
Objections to Turing Test
Objections to Turing Test
Foundations of AI
§ Philosophy: Aristotle, mechanistic views, materialism,
positivism, rationality.
§ Mathematics: algorithms, logic, formalization of
mathematics, incompleteness, decision theory.
§ Psychology: behaviorism, cognitive science.
§ Linguistics: grammars, syntax and semantics.
§ Computer Science: computers, software, theory
§ Others: neuroscience, economics, game theory.
Ready to play a Game ?
Tic-Tac-Toe
O
O
X
Another Game ?
Game of Nim
§ Robots § Game Playing
§ Chess-playing program § Machine Translation
§ Voice recognition system § Resource Scheduling
§ Expert systems (diagnosis,
§ Speech recognition system
advisory, planning, etc)
§ Grammer checker § Machine learning
§ Pattern recognition § Intelligent interfaces
§ Medial diagnosis
§ System malfunction rectifier
A Robot Colony
§ Problem Definition:
Intelligent Agents
A ∩ (A → B ) ⇒ B
Inductive Reasoning
§ Used to arrive at a general conclusion from a
limited set of facts
§ Eg:
§ Premise: Crows in Sri Lanka can fly
§ Premise: Crows in India can fly
§ Conclusion: Crows can fly
§ For a set of objects, X={a,b,c,d,...} if property P is true for
a, and if P is true for b and if P is true for c, then P is true for
all X
§ Not true always
Abductive Reasoning
Stench
PIT
Breeze
Hunter à PIT
Exploring the world of Wumpus
Knowledge Base:
Stench: No
PIT Breeze: No
Prize: No
OK: [ 1,1]
PIT: - Not in [1,1]
Wu: - Not in [1,1]
PIT
Inference:
OK: [2,1] and [1,2]
PIT: Not in [1,2], [2,1]
Wu: Not in [1,2], [2,1]
Action:
à Move to (1,2)
PIT
Exploring the world of Wumpus
Knowledge Base:
Stench: No
PIT Breeze: Yes
Prize: No
OK: [1,2], [2,1]
PIT: - Not in [1,1], [1,2], [2,1]
Wu: - Not in [1,1], [1,2], [2,1]
PIT
Inference:
Pit in [3,1] or [2,2] or [1,1]
à Pit in [3,1] or [2,2]
Wu: Not in [1,1],[2,2] and [3,1]
Actions:
§ à Move to (1,1)
PIT § à Move to [1,2]
Exploring the world of Wumpus
Knowledge Base:
Stench: Yes
PIT Breeze: No
Prize: No
OK: [1,2], [2,1]
PIT: Not in [1,1], [1,2], [2,1]
In [2,2] or [3,1] or both
PIT Wu: Not in [1,1], [1,2], [2,1], [2,2], [3,3]
Inference:
Wu: in [1,3], [2,2] or [1,1]
è Wu in [1,3]
Pit: not in [1,3], [2,2] or [1,1]
è Pit: In [3,1]
è [2,2] OK
Action:
Move to [2,2]
PIT
Exploring the world of Wumpus
Knowledge Base:
Stench: Yes
PIT Breeze: No
Prize: No
OK: [1,2], [2,1]
PIT: Not in [1,1], [1,2], [2,1]
In [2,2] or [3,1] or both
PIT Wu: Not in [1,1], [1,2], [2,1], [2,2], [3,3]
Inference:
Wu: in [1,3], [2,2] or [1,1]
è Wu in [1,3]
Pit: not in [1,3], [2,2] or [1,1]
è Pit: In [3,1]
PIT
Illustrative example: taxi driver
synapse
synapse axon
axon
nucleus
nucleus
cell
cellbody
body
dendrites
dendrites
The Structure of Neurones
• A neurone only fires if its input signal
exceeds a certain amount (the threshold) in
a short time period.
• Synapses vary in strength
– Good connections allowing a large signal
– Slight connections allow only a weak signal.
– Synapses can be either excitatory or inhibitory.
A Classic Artifical Neuron
ao+1
wj0
a1 wj1
wj2
a2
Sj f (Sj) Xj
wjn
an
Multilayer Perceptron
Output Values
Output Layer
Adjustable
Weights
Input Layer
Y
wn
net input to Y: xn
n
bias b is treated net = bweight
as the + x ifrom ∑
w i a special unit with
constant output 1. i =1
threshold related to Y
output
( x1 , ...... xn )
§ Decision region/boundary
n = 2, b != 0, θ = 0 x2 +
b + x1 w1 + x 2 w2 = 0 or -
w1 b
x 2a =line,
is − calledx1 −decision boundary, which partitions the planex
into two w 2 decision
w 2regions 1
b + x1 w1 + x 2 w 2 < 0
Case Studies
PC Based Fetal Monitoring Unit
Heart Beat wave shape
Neural Network
W1,1,1 W2,1,1
1 1 1
Z-1 W1,1,2 W2,1,2
W3,1,1
2 2 2
W3,2,1
Z-1
W3,3,1 Out put
3 3 3
. . .
. . .
. . . W3,10,1
Third layer
. . .
Z-1
15 10 10
Inputs First layer
(15 at a time) Second layer
Training the Neural Network
Training performances
Recognition
5
Heart rate:110 Base line:110 B.L. Var. :+1 Print
6
DECELERATION WARNING !
Example
Automated character recognition
C:\ocrsamp\janaka7.tif
C:\ocrsamp\janaka11.tif
Example
software to recognize handwritten
characters...
Methodology
Step 1: Taking the letters apart
Methodology
Step 2: Standardizing the letters
1. Skeltonizing
2. Normalizing
Methodology
Step 3: Extracting Features
16 features
Feature Space l
l
mm m
1 m
i
m
0.9 mm w
m w mmw
w wm m w an
m a
0.8 l n wd
w nmwu a a g
w aa a l
wl l
0.7 n x xnndux a a a a a h a a
a aa g
l d
ai i
du kd p
l aaa y a a a i a sg
w l dunuu xn u da n a zakl aaa aa
a z
ak a
xni y bu
n nni kudd a
nxn n nk
a
nay f d yka
ag g e ggsg g
g
iy du y
i a dh aaar da hd
z se e
Mean (F9..F16)
0.6 yh n nru
n hzudh
nu
xa e o
nvnn n n hr nk f
y xhlu avx ny a kyb qi o o pl g o pg egg ee
r u vfw n uvgu j uu nfnf q qgq oqoboo
oepb esp e s es
i y l
v
tn f vtq t t uu n
u uhri q tq kksoe ooebe q o o
e be s
bp e e es e
0.5 r ity lydlnq if ii t k kn zb tp
it lya o bo
oeoe
s
o kbe e
pep es ee
yy fprw
ri rirfrtu
jr yftzh
vtgal kl ayifllzi jo habhh krpz soc
oppoqe p b t
s
l r vo l ceepc
i w l i r r k
iv h h k l p kr t o e f s s c t
r
0.4 j aj a ris asr h or sso s cc
if jj j i i ks r r l t cc
i j r c
t r j i cc c
0.3 j i j j nt
c c
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Mean (F1..F8)
1
0.9
Feature Space i
a
0.8 a
u a a
aaaa a
u a aa
0.7 i
a i aa a aa aaa i a
a
a
u u aa a a aa a a
u u i uu a aaa a a a a e
i i u aa e e
Mean (F9..F16)
0.6 u uu u i a aaa a
u a a a e oo
i o ooe e e ee e
u u uu u oo oo oo e e e
i i i i ii uu u i oeo oooee ee e
eo e
e
0.5 a i oa o oo ee ee ee
oo
i i u a o ai ee o e
i i
0.4 i i a
o oo e
i ai a
i i
i i
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Mean (F1..F8)
Self Adaptive Artificial
Neural Network (SAANN)
Classifier
Character
Features
SAANN Character
EXPERIENCE
With 3 new users
With 1 user
However, practically
these waveforms
deviate from its pure
sinusoidal form due to
many reasons
Research Background
…equipment
causing PQ
disturbances
have increased dramatically...
…..The research has developed a novel Intelligent Power
Quality Monitor ( IPQM ), which is capable of…..
Detecting
Capturing
Classifying
….PQ disturbances
Locating
Analysing
….Harmonic sources
IPQMS
Bus 1 Bus 2 Bus N
50
DMU 4
Re (C ) ⋅ Vr (t ) + Im( C ) ⋅ Vr (t − )
40
30
T 2
20
e(t ) = v(t ) − 4
e(t)
0
10
0
C -1
-2
r= C
-10
-20 -3
-30 -4
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
where,
Voltage Signal T
T
C = ∫ v(t ) ⋅ Vr (t ) + j ⋅ v (t ) ⋅ Vr (t − ) ⋅ dt Extracted Disturbance
0
4
1
Voltage (p.u.)
-1
-2
0 0.2 0 .4 0 .6 0.8 1 1 .2 1.4
tim e ( s )
0 .5
Voltage (p.u.)
-0 .5
-1
-1 .5
0 0.2 0 .4 0 .6 0.8 1 1 .2 1.4
tim e ( s )
Example
2
Voltage (p.u.) 1
V(t)
0
-1
-2
0 0.2 0.4 0.6 0.8 1 1.2 1.4
tim e ( s )
0.5
Voltage (p.u.)
-0.5
-1
e(t) -1.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4
tim e ( s )
Example
2
Voltage (p.u.) 1
V(t)
0
-1
-2
0 0.2 0.4 0.6 0.8 1 1.2 1.4
tim e ( s )
0.5
Voltage (p.u.)
-0.5
-1
e(t) -1.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4
tim e ( s )
Example
1
V(t)
Voltage (p.u.)
-1
-2
0 0.2 0.4 0 .6 0.8 1 1 .2 1.4
time (s)
0.2
0.1
Voltage (p.u.)
-0.1
e(t) -0.2
-0.3
0 0.2 0.4 0 .6 0.8 1 1 .2 1.4
time (s)
Example
1
V(t)
Voltage (p.u.)
-1
-2
0 0.2 0.4 0 .6 0.8 1 1 .2 1.4
time (s)
0.2
0.1
Voltage (p.u.)
-0.1
e(t) -0.2
-0.3
0 0.2 0.4 0 .6 0.8 1 1 .2 1.4
time (s)
Distributed Monitoring Unit
Bus
Voltage Waveform V(t)
Distributed
Monitoring Disturbance Extraction Module
Unit
[ DMU ] Extracted Disturbance
waveform e(t)
Event Identification
Module ( EIM )
State Model
Steady Transition
State State
Captured
Event
Captured
Event Captured
Event
Intermediate
Intermediate Steady State
Transition State
Example
(a)
2
voltage (pu)
0
V(t)
-2
0 0.05 (b)
0.1 0.15 0.2
1
time (s)
voltage (pu)
0
e(t)
-1
0 0.05 (c)
0.1 0.15 0.2
2
time (s)
STATE
1.5
1
0 0.05 (d)
0.1 0.15 0.2
1
time (s)
Extracted
Voltage (pu)
event -1
0
V(t) -2
0 0.05 (b)
0.1 0.15 0.2
1
time (s)
voltage (pu)
0
e(t) -1
0 0.05 (c)
0.1 0.15 0.2
2
time (s)
STATE
1.5
1
0 0.05 (d)
0.1 0.15 0.2
1
time (s)
Voltage (pu)
Extracted0
event -1
0 0.005 0.01 0.015 0.02
wvaeform time (s)
Distributed Monitoring Unit
Bus
Voltage Waveform V(t)
Distributed
Monitoring Disturbance Extraction Module
Unit
[ DMU ] Extracted Disturbance
waveform e(t)
Under Normal
Voltage Voltage Voltage Condition
Sag Swell
SIMULATION
Biometric Recognition System
§ Facial Recognition
§ Voiceprint Recognition
§ Fingerprint Recognition
§ Retinal Patterns Recognition
§ DNA Recognition
ECG signal classification
Thank You !