Biomedical Engineering Reference
In-Depth Information
Machine Learning
The pattern-matching and pattern discovery components of data mining are often performed by
machine learning techniques. Machine learning isn't a single technology or approach, but
encompasses a variety of methods that represent the convergence of several disciplines, including
statistics, biological modeling, adaptive control theory, psychology, and artificial intelligence (AI).
Although many computer scientists consider the entire field of machine learning to be an outgrowth
of traditional statistical methods, biological modeling is clearly a source of several machine learning
approaches. These include genetic algorithms and neural networks. Similarly, adaptive control
theory, in which system parameters change dynamically to meet the current conditions, and
psychological theories, especially those regarding positive and negative reinforcement learning,
heavily influence machine learning methods. AI techniques, such as pattern matching through
inductive logic programming, are designed to derive general rules from specific examples. As
illustrated in
Table 7-3
, the spectrum of machine learning technologies applicable to data mining
includes inductive logic programming, genetic algorithms, neural networks, statistical methods,
Bayesian methods, decision trees, and Hidden Markov Models.
Table 7-3. Machine Learning Technologies and Their Applicability to Data-
Mining Methods.
Data-Mining Methods
Machine Learning
Technologies
Link
Analysis
Deviation
Detection
Classification Regression Segmentation
Inductive Logic
Programming
X
X
Genetic Algorithms
X
X
X
Neural Networks
X
X
X
Statistical Methods
X
X
X
X
X
Decision Trees
X
X
Hidden Markov
Models
X
Regardless of the underlying technology, most machine learning follows the general process outlined
in
Figure 7-6
. Input data are fed to a comparison engine that compares the data with an underlying
model. The results of the comparison engine then direct a software actor to initiate some type of
change. This output, whether it takes the form of a change in data or a modification of the underlying
model, is evaluated by an evaluation engine, which uses the underlying goals of the system as a
point of reference. Feedback from the actor and the evaluation engine direct changes in the model.
In this scenario, the goals can be standard patterns that are known to be associated with the input
data. Alternatively, the goals can be states, such as minimal change in output compared with the
system's previous encounter with the same data.
Figure 7-6. The Machine Learning Process.
