# Fuzzy Signatures

Document Sample

```					Fuzzy Signatures in
SARS

Student: Bai Qifeng
Client: Prof. Tom Gedeon

1
Problem of Fuzzy Theory
   A Major issue in fuzzy applications is how to create
fuzzy rules

   the number of rules have an exponential increase
with the number of inputs and terms.
| R | O (T k )

   At least one activated rule for every input.
e.g. 5 terms, 2 inputs => 25 rules
5 terms, 5 inputs => 3,125 rules

2
Fuzzy Signature
 Fuzzy signatures structure data into vectors
of fuzzy values, each of which can be a
further vector.

 Fuzzy signatures can be regarded as Special
multidimensional fuzzy data. Some of the
dimensions are interrelated in the sense that
they form sub-group of variables, which jointly
determine some feature on a higher level.

3
Fuzzy Signature

4
Fuzzy Signatures
 The relationship between higher and lower
levels is govern by fuzzy aggregations.

 Appropriate aggregations used in their child
signatures are not necessary identical.

 They can be changed based on expert
opinions and detailed circumstance.

5
Fuzzy Signatures in SARS
 The following scheme is of some daily symptom
signatures of patients:


            8am   
 8am            
       12 pm                       12 pm  
 f ever                      fever       
        4 pm                       4 pm   
                       main              
AS          8 pm      AS              8 pm   
        12am                       12am  
Cough                        Cough        
         9 pm  
                       9 pm   
    Nausea                           Nausea  
                          Secondary          
      Sore                           Sore  
                        
6
Automatically Construct Fuzzy
Signature
 Searches the features of data structure,
classify data based on their relevance and
cluster the high relevant data.

 With clustering, the known situation can be
used to build the model and then it can be
used to another situation where it is not
known.

7
Fuzzy Clustering
    Hierarchical Method
Creating a cluster tree.

   Objective Function Method
Solving problems about fuzzy boundary in
evaluating relevance of objects

8
Data Pretreatment
 Standardizing raw data

 Outlier
   Two-stage method
   Scatter Plot

 Missing data

9
Conclusion
 Choosing proper cluster algorithm and
appropriate data pretreatment, try to find the
appropriate fuzzy signatures

 After constructing the fuzzy signature, with
aggregations, we can effectively reduce the
number of rules in this fuzzy system.

10

```
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 3 posted: 6/4/2012 language: pages: 10