Transcript march29 - College of Computer and Information Science

CS 4100 Artificial Intelligence
Prof. C. Hafner
Class Notes March 27, 2012
Term Project Presentations
• Homework 6 is due Tuesday in class (hard copy)
• We need 4 teams to volunteer to make presentations
on April 12 !!
• The other 5 teams will make presentations on April 17
(last day)
• Each presentation will be strictly limited to 15 minutes,
with 3 minutes for discussion/questions.
• Make sure your slides/demos load immediately – we
do not have time to wait for Google Docs exploration.
Supervised Learning (cont.)
• Decision Tree Learning (actually, classification
learning) – return for further discussion
• We also consider techniques for evaluating
supervised learning systems
• Perceptrons/Neural Nets
• Naïve Bayes Classifiers
• April 3, 5, 10: finish ML, introduce NLP
• http://www.cis.temple.edu/~giorgio/cis587/readings
/id3-c45.html#1.
ID3 and C4.5 Golfing Example: Attributes
Decision: Play or Don’t Play
ID3 and C4.5 Golfing Example: Training Data
Decision: Play or Don’t Play
Stock Market Example
Table of Entropy values
• http://usl.sis.pitt.edu/trurl/log-table.html
Review the Algorithm
• In the case of our golfing example, for the attribute Outlook we
have
• Info(Outlook,T) = 5/14*I(2/5,3/5) + 4/14*I(4/4,0) +
5/14*I(3/5,2/5) = 0.694 Consider the quantity Gain(X,T) defined
as
• Gain(X,T) = Info(T) - Info(X,T) This represents the difference
between the information needed to identify an element of T and
the information needed to identify an element of T after the value
of attribute X has been obtained, that is, this is the gain in
information due to attribute X.
• In our golfing example, for the Outlook attribute the gain is:
• Gain(Outlook,T) = Info(T) - Info(Outlook,T) = 0.94 - 0.694 = 0.246.
If we instead consider the attribute Windy, we find that
Info(Windy,T) is 0.892 and Gain(Windy,T) is 0.048. Thus Outlook
offers a greater informational gain than Windy.
C4.5 Extension Example 1
Notice that in this example two of the attributes have continuous ranges,
Temperature and Humidity. ID3 does not directly deal with such cases.
We can deal with the case of attributes with continuous ranges as follows:
Say that attribute Ci has a continuous range. We examine the values for this
attribute in the training set. Say they are, in increasing order, A1, A2, ..,
Am. Then for each value Aj, j=1,2,..m, we partition the records into those
that have Ci values up to and including Aj, and those that have values
greater than Aj. For each of these partitions we compute the gain, or gain
ratio, and choose the partition that maximizes the gain. This makes Ci a
Boolean (or binary) attribute.
In our Golfing example, for humidity, if T is the training set, we determine
the information for each partition and find the best partition at 75. Then
the range for this attribute becomes {<=75, >75}. Notice that this method
involves a substantial number of computations.
C4.5 Extension Example 2
ID3 and C4.5
• ID3 algorithm (we learned last time) is important not
because it summarizes what we know, i.e. the training
set, but because we hope it will classify correctly new
cases. Thus when building classification models one
should have both training data to build the model and
test data to verify how well it actually works.
• C4.5 is an extension of ID3 that accounts for unavailable
values, continuous attribute value ranges, pruning of
decision trees, rule derivation, and so on.
Perceptrons and Neural Networks:
Another Supervised Learning Approach
Perceptron Learning (Supervised)
•
•
•
•
Assign random weights (or set all to 0)
Cycle through input data until change < target
Let α be the “learning coefficient”
For each input:
– If perceptron gives correct answer, do nothing
– If perceptron says yes when answer should be no,
decrease the weights on all units that “fired” by α
– If perceptron says no when answer should be yes,
increase the weights on all units that “fired” by α
Naive Bayes Classifiers:
Our next example of machine learning
• A supervised learning method
• Making independence assumption, we can explore a
simple subset of Bayesian nets, such that:
• It is easy to estimate the CPT’s from sample data
• Uses a technique called “maximum likelihood
estimation”
– Given a set of correctly classified representative
examples
– Q: What estimates of conditional probabilities maximize
the likelihood of the data that was observed?
– A: The estimates that reflect the sample proportions
# Juniors
were Juniors and
# Juniors
were Non-Juniors
# Non-Juniors
Naive Bayes Classifier
with multi-valued variables
Major: Science, Arts, Social Science
Student characteristics:
Gender (M,F),
Race/Ethnicity (W, B, H, A)
International (T/F)
What do the conditional probability tables look like??
Theoretical Foundation and Application to Text
Classification - thanks Prof. Daphne Koller at Stanford