Abstract | ||
---|---|---|
The segmentation of a signal based on a piecewise polynomial model is reexamined here in light of the recent advances in applying the MDL principle to the exponential density family. The critical case of discriminating between short segments is handled very efficiently by using the exact MDL formula for small sample size, whereas MAP or AIC methods result in drastic over-segmentation. The simulation result for ECG segmentation shows that the piecewise linear approximation obtained with the proposed method preserves well the location of the QRS complex. |
Year | DOI | Venue |
---|---|---|
2000 | 10.5281/zenodo.37601 | EUSIPCO |
Keywords | Field | DocType |
indexes | Exponential function,Segmentation,Piecewise linear approximation,Digital signal,Algorithm,Sample size determination,Piecewise,Mathematics,Polynomial and rational function modeling | Conference |
ISBN | Citations | PageRank |
978-952-1504-43-3 | 1 | 0.70 |
References | Authors | |
0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ciprian Doru Giurcaneanu | 1 | 43 | 12.44 |
Tabus, Ioan | 2 | 35 | 4.68 |
Jorma Rissanen | 3 | 1665 | 798.14 |