Download Diapositiva 1

Intro Machine
Learning
Strata 2 – Teknologi Informasi
STTS
Computer Science Area’s
•
•
•
•
•
•
•
•
•
Artificial intelligence
Web Programming
Algorithm and Data Structure
Computer Architecture
Computer Graphics
Software Engineering
Database and Operating System
Theory of Computation
etc
Referensi:
https://en.wikipedia.org/wiki/Outline_of_computer_science
http://www.cs.cornell.edu/Info/Department/Ugrad/Subareas.html
Computer Science :
definition
Computer Science is the study of computers and
computational systems.
Unlike electrical and computer engineers,
computer scientists deal mostly with software
and software systems; this includes their theory,
design, development, and application.
Referensi:
http://undergrad.cs.umd.edu/what-computer-science
Artificial Intelligence
: definition
Artificial Intelligence (AI) is an area of computer
science that emphasizes the creation of
intelligent machines that work and react like
humans.
Part of Artificial
Intelligence
• Expert System / KBS
• Machine Learning
• Solution Search
• Computer Vision
• Digital Image Understanding
• Digital Image Processing
• Game Playing
• Voice Recognition
• Speech Recognition
• Robotic
Machine Learning
• Machine Learning: Field of study that gives
computers the ability to learn without being
explicitly programmed Arthur Samuel (1959).
Machine Learning
Problem
Learning Process
Data
Term:
• Supervised
• Unsupervised
• Discreet
• Continous
Supervised
Unsupervised
x2
vs
x2
x1
x1
Example 1
Price ($) 400
in 1000’s
Linear
300
?
Polynomial
200
100
0
0
500
1000
1500
2000
750
By Learning Method : Supervised
Regression
By Data : Continous
2500
Size in feet2
Example 2
Breast Cancer : Malignant OR Benign
1(Y)
Malignant?
0(N)
Tumor Size
By Learning Method : Supervised
Classification
By Data : Discreet  YES or NO
Other Example 3
- Clump Thickness
- Uniformity of Cell Size
- Uniformity of Cell Shape
…
Age
Tumor Size
By Learning Method : Supervised
Classification
By Data : Discreet  YES or NO
Exercise
• Problem 1:
You have a large inventory of identical items. You want to
predict how many of these items will sell over the next 3
months.
• Problem 2:
You’d like software to examine individual customer accounts,
and for each account decide if it has been hacked/
compromised.
Answer
• Treat problem 1 as a regression problem
• Treat problem 2 as a classification problem
Classification Example
Training
Training
Images
Image
Features
Training
Labels
Classifier
Training
Trained
Classifier
Cont..
Testing
Image
Features
Trained
Classifier
Prediction
Outdoor
Classification
• Given some set of features with corresponding labels
• Learning a function to predict the labels from the features
• Training labels dictate that two examples are the same or
different, in some sense
• Features and distance measures define similarity
• Classifiers try to learn weights or parameters for features and
distance measures so that feature similarity predicts label
similarity
Clustering Example
Cont..
Exercise
(supervised OR unsupervised)
 Given email labeled as spam/not spam, learn a spam filter.
 Given a set of news articles found on the web, group them
into set of articles about the same story.
 Given a database of customer data, automatically discover
market segments and group customers into different market
segments.
 Given a dataset of patients diagnosed as either having
diabetes or not, learn to classify new patients as having
diabetes or not.
Intro to NN
Neural Network (NN)
• Origins: Algorithms that try to mimic the
brain.
• Was very widely used in 80s and early 90s;
popularity diminished in late 90s.
• Recent resurgence: State-of-the-art technique
for many applications
NN Representation
NN Model
Neural Network has over 20's models, each of
which can be distinguished by:
• Architecture
• How to learn
Architecture
• Full Connected Graph
• Feed Forward
Term
•
•
: Node/ Vertex
: Edge
Layer
Mc Culloch - Pitt
• Node input disebut X
X1
• Bobot disebut W (weight)
W1
X2
W2
W3
X3
.
.
. Wn
.
.
Xn
• Bobot layer 1 disebut V
Y1
• Bobot layer 2 disebut W
(kebanyakan hanya sampai 2 layer)
• Perubahan V disebut ΔV
• Perubahan W disebut ΔW
• Output disebut Y
Sum of Product
• Penjumlahan Hasil Perkalian
 ( x1  w1)  ( x2  w2)  ( x3  w3)  ....  ( xn  wn)
n
Y
in
  ( xi  wi )
i 1
Fungsi Aktivasi
• Binary Hard Threshold
1,

f ( x)  
0
if
x 
Contoh 1
• Berikut ini adalah contoh bahwa bobot merepresentasikan
pengetahuan, sehingga jika diberi pertanyaan dapat diketahui
jawabannya.
X1
X1
X2
Y1
0
0
0
0
1
0
1
0
0
1
1
1
1
1
Y1
X2
Dengan ϴ = 2
1,

f ( x)  
0
if
x 
Contoh 2
X1
2
2
Y1
X2
Dengan ϴ = 2
1,

f ( x)  
0
if
x 
X1
X2
Y1
0
0
0
0
1
1
1
0
1
1
1
1
Contoh 3
Pikirkan berapakah bobot untuk :
X
X1 AND (NOT X2)
1
2
-1
Y1
X2
Dan
X2 AND (NOT X1)
Dengan ϴ = 2
X1
-1
2
X2
Y1
Contoh 4
• Pikirkan berapakah bobot untuk menangani:
X1 XOR X2
Sama artinya dengan...
(X1 AND (NOT X2)) OR (X2 AND (NOT X1))
X1
2
Z1
-1
-1
X2
2
2
2
Z2
Y1
Dengan ϴ = 2
Kasus
• Linear Separable
x1
x1
x1
1
1
1
x2
x2
1
AND
x2
1
1
AND NOT
OR
• Non Linear Separable
x1
1
x2
1
XOR
Summary
• Different weights can save different
knowledge
• The more number of nodes, the knowledge
that is stored will be more complex