Biomedical Engineering Reference
In-Depth Information
Equating the partial derivatives with respect to the coefficients as before allows us to
construct a set of simultaneous equations that can be solved to find the coefficients
n
n
n
n
z
i
=
a
1
+
b
x
i
+
c
y
i
i
=
1
i
=
1
i
=
1
i
=
1
n
n
n
n
x
i
x
i
z
i
=
a
x
i
+
b
+
c
x
i
y
i
(5.70)
i
=
1
i
=
1
i
=
1
i
=
1
n
n
n
n
y
i
y
i
z
i
=
a
y
i
+
b
x
i
y
i
+
c
i
=
1
i
=
1
i
=
1
i
=
1
When working with categories of outputs rather than numbers, classification algorithms
are a better choice than the regression methods so far discussed.
5.6.2 Data Mining
Data mining has been defined as the process of extracting patterns from very large data sets
by using a combination of statistical and machine learning techniques in conjunction with
database management. Typically, the focus of data mining is primarily on data storage,
integration, and retrieval, with the analysis functions being of secondary importance.
5.6.3 Machine Learning
Machine learning (ML) uses computationally powerful methods to learn very complex
nonparametric (or quasi-parametric) models of the data. The models used are often close
to human representations, and the process is one of learning or improved performance
with “experience.”
ML techniques include the following:
• Artificial intelligence
• Decision/classification trees
• Clustering
• Support vector machines
• Expert systems
• Neural networks
• Genetic algorithms
• Bayesian (belief) networks
• Bayesian statistics and nets
ML can be divided into two broad categories: supervised and unsupervised learning.
In supervised learning, the algorithms are presented with a number of examples. This
allows the algorithm to learn to predict the correct output corresponding to previously
seen inputs as well as previously unseen ones. In unsupervised learning the algorithm is
required to generate its own categories of outputs.
