Recognition of facial expressions - Knowledge Based Systems ...
Recognition of facial expressions - Knowledge Based Systems ...
Recognition of facial expressions - Knowledge Based Systems ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
For the case that P h ) = P(<br />
h ) , a simplification can be further done by choosing the<br />
(<br />
i<br />
j<br />
Maximum likelihood (ML) hypothesis:<br />
h<br />
ML<br />
= arg<br />
max P(<br />
D / h )<br />
hi ∈ H<br />
i<br />
The Bayesian network is a graphical model that efficiently encodes the joint probability<br />
distribution for a given set <strong>of</strong> variables.<br />
A Bayesian network for a set <strong>of</strong> variables X = X ,..., X } consists <strong>of</strong> a network structure<br />
{ 1 n<br />
S that encodes a set <strong>of</strong> conditional independence assertions about variables in X , and a<br />
set P <strong>of</strong> local probability distributions associated with each variable. Together, these<br />
components define the joint probability distribution for X . The network structure S is a<br />
directed acyclic graph. The nodes in S are in one-to-one correspondence with the<br />
variables X . The term<br />
and<br />
pai<br />
to denote the parents <strong>of</strong> node<br />
those parents.<br />
X<br />
i<br />
is used to denote both the variable and the corresponding node,<br />
X<br />
i<br />
in S as well as the variables corresponding to<br />
Given the structure S, the joint probability distribution for X is given by:<br />
p(<br />
x)<br />
=<br />
Equation 1<br />
n<br />
∏<br />
i=<br />
1<br />
p(<br />
x i<br />
| pa i<br />
)<br />
The local probability distributions P are the distributions corresponding to the terms in<br />
the product <strong>of</strong> Equation 1. Consequently, the pair (S;P) encodes the joint distribution<br />
p(x).<br />
The probabilities encoded by a Bayesian network may be Bayesian or physical. When<br />
building Bayesian networks from prior knowledge alone, the probabilities will be<br />
Bayesian. When learning these networks from data, the probabilities will be physical (and<br />
their values may be uncertain).<br />
Difficulties are not unique to modeling with Bayesian networks, but rather are common<br />
to most approaches.<br />
As part <strong>of</strong> the project several tasks had to be fulfilled, such as:<br />
- 33 -