Databases Reference
In-Depth Information
Fig. 25.
Example of weight decay. Here
r
was set to 2. The left picture shows the initial activations
and similarity scores while the right picture shows the results after 3 iterations. Note that for the
sake of completeness the weights of the edges were not set to 0 when they reached
.
!" #
$#
%&#
!" #
$#
%&#
!" #
$#
%&#
!" #
$#
%&#
=
=
(a) Population
20 individuals.
(b) Population
100 individuals.
Fig. 26.
F-score and runtime on the ACM-DBLP dataset. f(X) stands for the F-score achieved by
algorithm X, while d(X) stands for the total duration required by the algorithm.
OAEI 2010 benchmark
20
. The real-world datasets consisted of the ACM-DBLP and
Abt-Buy datasets, which were extracted from websites or databases [76]
21
. Given that
genetic programming is non-deterministic, all results presented below are the means
of 5 runs. Each experiment was ran on a single thread of a server running JDK1.7
on Ubuntu 10.0.4 and was allocated maximally 2GB of RAM. The processors were
2.0GHz Quadcore AMD Opterons. An excerpt of the results is shown in Figure 26.
While the results show that COALA outperform EAGLE, it remains unclear whether
WD or CL is the best approach to achieving a faster convergence towards the optimal
solution.
