| |
 |
|
|
|
| |
 |
|
| |
|
|
| |
| |
Early stopping
A method of optimization in which the objective used is not
the real goal, and optimization is stopped when another measure
of fit starts to rise. This may be critically dependent on the
starting value chosen.
|
|
|
|
| |
|
|
| |
| |
Editing
Methods of reducing the training set for sue by nearest-neighbour
methods
|
|
|
|
| |
|
|
| |
| |
Efficiency
A statistical term, measuring the performance of estimators.
Unless stated otherwise, efficiency is a measure of 1/(n Ąż variance
in samples of size n) (for large n).
|
|
|
|
| |
|
|
| |
| |
Eigen Decomposition
of a real symmetric matrix A. This is an orthonormal matrix
C and a non-negative diagonal matrix D such that A=CDC .
|
|
|
|
| |
|
|
| |
| |
EM Algorithm
A device to construct algorithms for maximum likelihood and
MAP estimators
|
|
|
|
| |
|
|
| |
| |
Equivariance
An invariant procedure is unchanged under a transformation;
an equivariant procedure transforms its answer. For example,
if is a parameter and
|
|
|
|
| |
|
|
| |
| |
Estimator
A rule to assign a parameter value to a set of observations.
The value assigned is called an estimate, and the distinction
between estimators and estimators is not always observed.
|
|
|
|
| |
|
|
