Conditional

Fuzzy C Means
A fuzzy clustering approach for
mining event-related dynamics
Christos N. Zigkolis

Contents
The problem
Our approach
Fuzzy Clustering
Conditional Fuzzy Clustering
Graph-Theoretic Visualization techniques
The experiments and the datasets
Applications
Future Work
Conclusions
2

Aristotle University of Thes

saloniki

The problem
Visualizing the variability of MEG responses
understanding the single-trial variability

Describe the single-trial (EEG) variability in the presence of artifacts

make single-trial analysis robust, robust prototyping

Aristotle University of Thes

saloniki

Our approach
criteria

CONDITIONAL

grades

content constraints

0 or 1

FUZZY
partial membership

CLUSTERING
creating clusters
4

Aristotle University of Thes

saloniki

[0, 1]

Clustering
Fuzzy Clustering
Every
one
cluster
EveryPattern
Patternto
toonly
every
cluster
with partial membership

Aristotle University of Thes

saloniki

Fuzzy C Means
XdataNxp

U membership matrix

u11 u1N

uC1 uCN

Centroids

V [V1 , V2 ,..., VC ]

Objective function

| Q(t ) Q (t 1) |

CONTINUE STOP

Iterative procedure
6

Aristotle University of Thes

saloniki

FCM 2D Example
compact groups

spurious patterns

FCM sensitivity to
noisy data

Aristotle University of Thes

saloniki

Conditional Fuzzy Clustering

The presence of Condition(s)

mark

Pattern

Condition(s)

Aristotle University of Thes

saloniki

Conditional Fuzzy C Means

F [ f1 , f 2 ,..., f N ]
scaled to [0, 1]

<Xdata, F> CFCM <U, Centroids>

F affects the computations of U matrix and consequently the centroids.

Aristotle University of Thes

saloniki

FCM VS CFCM
FCM

uij

uij

uij

uij

10

uij

Fk

VS
uij

CFCM

uij

Aristotle University of Thes

saloniki

uij

Graph-theoretic
Visualization Techniques
Topology Representing Graphs
Build a graph G [C x C]

Topological relations between prototypes

Gij corresponding to the strength of connection between prototypes O i and Oj

Computation of the graph G
- For each pattern find the nearest prototypes and increase the
corresponding values in G matrix
- Simple elementwise thresholding Adjacency Matrix A
A: a link connects two nearby prototypes only when they are natural
neighbors over the manifold
11

Aristotle University of Thes

saloniki

12

Aristotle University of Thes

saloniki

Graph-theoretic
Visualization Techniques
Compute the G graph via CFCM results
Apply CFCM algorithm: (O, U) = CFCM(X, Fk, C)
Build

U ' [uij' ]CxN , such that uij' uij . (uij )

Compute G = U.UT

13

FCG: Fuzzy Connectivity Graph

Aristotle University of Thes

saloniki

Graph-theoretic
Visualization Techniques
Minimal Spanning Tree

MST-ordering

5
7

3
2

14

root

Aristotle University of Thes

saloniki

Minimal Spanning Tree with MST-ordering

15

Aristotle University of Thes

saloniki

Graph-theoretic
Visualization Techniques
Locality Preserving Projections
Dimensionality Reduction technique

Rp Rr r<p

Linear approach MDS, LE, ISOMAP

- generalized eigenvector problem

- use of FCG matrix

- select the first r eigenvectors and tabulate them (Apxr matrix)

P = [pij]Cxr = OA
Alternative to PCA: different criteria, direct entrance of a new point into the
subspace
16

Aristotle University of Thes

saloniki

17

Aristotle University of Thes

saloniki

The experiments

Magnetoencephalography

18

Electroencephalography

+ 197 single trials

110 single trials

+ control recording

Online outlier rejection

Aristotle University of Thes

saloniki

The datasets
Feature Extraction
MEG

EEG

pT
V

msec

X_data [197 x p1], p:number of features

19

msec

X_data [110 x p2], p:number of features

Aristotle University of Thes

saloniki

Applications (MEG)
Exploit the background noise for better clustering
MEG single trials

20

+ Spontaneous activity as a auxiliary set of signals

Exploit the distances

Aristotle University of Thes to extract the grades
saloniki

21

Aristotle University of Thes

saloniki

Applications (EEG)
Robust Prototyping

Elongate the possible outliers

from the clustering procedure

Find the distances from the

nearest neighbors and compute
the grades for every pattern

22

Aristotle University of Thes

saloniki

FCM
23

Aristotle University of Thes

saloniki

CFCM
24

Aristotle University of Thes

saloniki

Future Work
Knowledge-Based Clustering Algorithms
conditional fuzzy clustering
wavelet transform

horizontal collaborative clustering

wavelet transform

25

Aristotle University of Thes

saloniki

Conclusions
Through the proposed methodology
exploit the presence of noisy data
elongate the outliers from the clustering procedure

Graph-Theoretic Visualization Techniques

study the variability of brain signals
study the relationships between clustering results

Paper submitted
Using Conditional FCM to mine event-related dynamics
26

Aristotle University of Thes

saloniki