slide 9.1 9 symbol-based machine learningmhtay/itec480/lecture/luger_ch09a.pdfmachine learning: 9...
TRANSCRIPT
MACHINE LEARNING: SYMBOL-BASED9
9.0 Introduction
9.1 A Framework for Symbol-based Learning
9.2 Version Space Search
9.3 The ID3 Decision Tree Induction Algorithm
9.4 Inductive Bias and Learnability
9.5 Knowledge and Learning
9.6 Unsupervised Learning
9.7 Reinforcement Learning
9.8 Epilogue and References
9.9 Exercises
Slide 9.1
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.2
Figure 9.1: A general model of the learning process.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.3
Figure 9.2: Examples and near misses for the concept “arch.”
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.4
Figure 9.3: Generalization of descriptions to include multiple examples.
(continued on next slide)
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.5
Figure 9.3: Generalization of descriptions to include multiple examples.(continued from previous slide)
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.6
Figu
re 9
.4:
Spe
cial
izat
ion
of a
des
crip
tion
to e
xclu
de
a ne
ar m
iss.
In 9
.4c
we
add
cons
train
ts to
9.
4a s
o th
at it
can
’t m
atch
with
9.4
b.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.7
Figure 9.5: A concept space.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.8
Defining specific to general search, for hypothesis set S as:
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.9
In this algorithm, negative instances lead to the specialization of candidate concepts; the algorithm uses positive instances to eliminate overly specialized concepts.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.10
Figure 9.6: The role of negative examples in preventing overgeneralization.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.11
Figure 9.7: Specific to general search of the version space learning the concept “ball.”
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.12
The algorithm specializes G and generalizes S until they converge on the target concept. The algorithm is defined:
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.13
Figure 9.8: General to specific search of the version space learning the concept “ball.”
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.14
Figure 9.9: The candidate elimination algorithm learning the concept “red ball.”
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.15
Figure 9.10: Converging boundaries of the G and S sets in the candidate elimination algorithm.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.16
Figure 9.11: A portion of LEX’s hierarchy of symbols.
A R T I F I C I A L I N T E L L I G E N C E: Structure and Strategies for Complex Problem Solving, 4th Edition George F. Luger © 2002 Addison Wesley
Slide 9.17
Figure 9.12: A version space for OP2, adapted from Mitchell et al. (1983).