Database Reference
In-Depth Information
395
392
202
0
0
0
0
0
0
0
200
151
82
fast ranking
C
=
D
= 5
time: 9.43 sec
fast ranking
C
=
D
= 2
time: 2.18 sec2
fast ranking
C
=
D
= 1.5
time: 1.09 sec
(a) Search for twelve-leg patterns of air and sea temperatures.
233
206
200
0
0
0
0
200
0
200
0
200
fast ranking
C
=
D
= 5
time: 1.58 sec
fast ranking
C
=
D
= 2
time: 1.06 sec
fast ranking
C
=
D
= 1.5
time: 0.61 sec
(b) Search for nine-leg patterns of wind speeds.
390
390
382
382
244
0
0
0
0
200
0
89
0
35
35
fast ranking
C
=
D
= 5
time: 0.31 sec
fast ranking
C
=
D
= 2
time: 0.10 sec
fast ranking
C
=
D
= 1.5
time: 0.05 sec
(c) Search for electroencephalogram patterns with twenty legs.
Fig. 15. Retrieval of weather and electroencephalogram patterns. The horizontal axes
show the similarity ranks assigned by the fast algorithm, and the vertical axes are the
exhaustive-search ranks.
whose compressed representation is similar to the compressed pattern.
The experiments have shown the effectiveness of this technique for index-
ing of stock prices, weather data, and electroencephalograms. We plan to
apply it to other time-series domains and study the factors that affect its
effectiveness.
We are working on an extended version of the compression procedure,
which will assign different importance levels to the extrema of time series,