next up previous contents
Next: INTRODUCTION Up: No Title Previous: LIST OF ABBREVIATIONS

LIST OF SYMBOLS

$\mbox{\boldmath$x$}$ input vector
$\mbox{\boldmath$m$}_i$ ith reference vector
c index of the closest $\mbox{\boldmath$m$}_i$ to $\mbox{\boldmath$x$}$
c' index of the second closest $\mbox{\boldmath$m$}_i$ to $\mbox{\boldmath$x$}$
N number of reference vectors
t discrete time index
hci neighborhood function in SOM
$\alpha$ learning rate
r neighborhood radius in SOM
D dimension of the input vectors
M dimension of the SOM display
$\mbox{\boldmath$1$}_r$ vector of zeros except the rth component is 1
$\mbox{\boldmath$I$}_i$ location of the ith SOM unit on the SOM grid
$p(\mbox{\boldmath$x$})$ probability density function of $\mbox{\boldmath$x$}$
$\epsilon$ stabilizing constant factor for LVQ3
Ci ith class in LVQ
M number classes in LVQ
$g_i(\mbox{\boldmath$x$})$ discriminant function of the ith class
$d_k(\mbox{\boldmath$x$})$ misclassification measure for $\mbox{\boldmath$x$} \in C_k$
$\mbox{\boldmath$O$}$ sequence observation vectors of a system
q sequence of states a system undergoes
N number of states in an HMM system
aij transition probability from state i to state j
$b_i(\mbox{\boldmath$O$}_t)$ output probability density model of state i
$\pi_i$ starting probability of state i
$\lambda$ description of an HMM
$\mbox{\boldmath$\mu$}_{jm}$ mean vector of mth Gaussian in jth codebook
$\mbox{\boldmath$c$}_{im}$ weight of mth Gaussian for state i
K number of best matching units used in density approximations
d average quantization error


next up previous contents
Next: INTRODUCTION Up: No Title Previous: LIST OF ABBREVIATIONS
Mikko Kurimo
11/7/1997