Embed Size (px)
Citation preview
1
AI and Machine Learning
FSEM 101_02 The Rise of Artificial Intelligence
The Role of Data
Data possess a great power and hold valuable
insights
Data-driven decision making process
Human overwhelmed by the sheer amount of data
to make sense of it all
Must rely on machines (computers) to carry out the
task automatically
Recognizing patterns and making inferences
2
A A A A A A A A A A A A A A A
A A A A A A A A A A A A A
A A A A A A A A A A A A A A
A A A A A A A A A A A A A
A A A A A A A A A A A A A
A A A A A A A A A A A A A
What is Machine Learning?
A field of study that gives computers the ability to learn
without being explicitly programmed (Arthur Samuel,
1959)
In 1998, Tom Mitchell of CMU described a well-posed
learning problem:
A computer program is said to learn from experience E
with respect to some task T and some performance
measure P, if its performance on T, as measured by P,
improves with experience E
3
Example: (Taken from Andrew Ng’s lecture on Machine Learning)
Suppose your e-mail program watches which e-mails you do or do not mark as spam, and based on that learns how to
better filter spam.
Experience E - watching you label e-mails as spam or not spam
Task T - classifying e-mails as spam or not spam
Performance P - the number or fraction of e-mails correctly classified as spam or not spam
Sources: intellspot.com/unsupervised-vs-supervised-learning/
4
Supervised Learning
Learns patterns from labeled data to predict future output
Labeled data data tagged with the right answer
Patterns can be much more complex than those found in unsupervised learning
Requires many examples of output for algorithm training
Typical task: classification, regression, ranking, structured prediction
Algorithms: nearest neighbors, decision trees, artificial neural networks, support vector machines, etc.
Nearest Neighbor Pattern Classification
5
Unsupervised Learning
Train computer to look for interesting patterns or
structures or hidden relationships in data without
any examples of output
Cannot demonstrate specific types of output
Best known as ‘Data Mining’
Typical task: Clustering (k-means, GMMs),
predictive modeling, summarization, visualization
6
K-means Clustering
Find a set of K clusters that best describe the data
Each cluster defined by a centroid
Instances belong to cluster with closest centroid
Use the Expectation-maximization algorithm
The goal of the algorithm is to minimize the cluster
scatter
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4ANEMIA PATIENTS AND CONTROLS
Red Blood Cell Volume
Red B
lood C
ell
Hem
oglo
bin
Concentr
ation
7
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 1
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 3
8
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 5
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 10
9
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 15
3.3 3.4 3.5 3.6 3.7 3.8 3.9 43.7
3.8
3.9
4
4.1
4.2
4.3
4.4
Red Blood Cell Volume
Re
d B
loo
d C
ell H
em
og
lob
in C
on
ce
ntr
atio
n
EM ITERATION 25
10
Applications
Spam Detection & Filtering
Human language Technology (speech, text,
gaze & gestures)
Image Recognition & Segmentation (face, hand-
writing, scene analysis, etc.)
Data Mining (consumer service
recommendation system, medical diagnosis )
Applications
Navigation system (autonomous robot, unmanned vehicle, computer vision)
Climatology (modeling & forecasting)
Bioinformatics (gene sequencing, medical diagnosis,….)
Adaptive control theory
Natural language Processing
11
12
13
