Acronym: DD_DTWType: Whole SeriesYear: 2014Publication: Knowledge-Based Systems

Description: Górecki and Łuczak describe an approach for using a weighted combination of raw series and first-order differences for NN classification with either the Euclidean distance or full-window DTW. They find the DTW distance between two series and the two differenced series. These two distances are then combined using a weighting parameter $\alpha$ (See Algorithm 4). Parameter $\alpha$ is found during training through a leave-one-out cross-validation on the training data. This search is relatively efficient as different parameter values can be assessed using pre-computed distances.

An optimisation to reduce the search space of possible parameter values is proposed. However, we could not recreate their results using this optimisation. We found that if we searched through all values of $\alpha$ in the range of $[0,1]$ in increments of 0.01, we were able to recreate the results exactly. Testing is then performed with a 1-NN classifier using the combined distance function given in Algorithm 4.
Source Code: Derivative DTW Code
Published Results:Recreated Results:

This algorithm doesn't have any published results.

Recreated
Dataset:Result:
Adiac0.6682
ArrowHead0.8381
Beef0.5567
BeetleFly0.8110
BirdChicken0.8405
Car0.7315
CBF0.9930
ChlorineConcentration0.7002
CinCECGtorso0.7309
Coffee0.9861
Computers0.7246
CricketX0.7613
CricketY0.7419
CricketZ0.7731
DiatomSizeReduction0.9578
DistalPhalanxOutlineCorrect0.7574
DistalPhalanxOutlineAgeGroup0.7333
DistalPhalanxTW0.6045
Earthquakes0.7079
ECG2000.8168
ECG50000.9261
ECGFiveDays0.7514
ElectricDevices0.7755
FaceAll0.9453
FaceFour0.8458
FacesUCR0.8979
FiftyWords0.7479
Fish0.9211
FordA0.7034
FordB0.7215
GunPoint0.9545
Ham0.6750
HandOutlines0.8681
Haptics0.3774
Herring0.5275
InlineSkate0.5525
InsectWingbeatSound0.3430
ItalyPowerDemand0.9199
LargeKitchenAppliances0.7933
Lightning20.8151
Lightning70.6937
Mallat0.9505
Meat0.9685
MedicalImages0.7375
MiddlePhalanxOutlineCorrect0.7295
MiddlePhalanxOutlineAgeGroup0.5765
MiddlePhalanxTW0.5106
MoteStrain0.8211
NonInvasiveFatalECGThorax10.8137
NonInvasiveFatalECGThorax20.8803
OliveOil0.8500
OSULeaf0.8845
PhalangesOutlinesCorrect0.7555
Phoneme0.2598
Plane0.9996
ProximalPhalanxOutlineCorrect0.8154
ProximalPhalanxOutlineAgeGroup0.7713
ProximalPhalanxTW0.7367
RefrigerationDevices0.6029
ScreenType0.5615
ShapeletSim0.6326
ShapesAll0.8621
SmallKitchenAppliances0.6397
SonyAIBORobotSurface10.8054
SonyAIBORobotSurface20.8630
StarlightCurves0.9654
Strawberry0.9560
SwedishLeaf0.8961
Symbols0.9556
SyntheticControl0.9906
ToeSegmentation10.7418
ToeSegmentation20.8280
Trace0.9997
TwoLeadECG0.9456
TwoPatterns1.0000
UWaveGestureLibraryX0.7743
UWaveGestureLibraryY0.7099
UWaveGestureLibraryZ0.6951
UWaveGestureLibraryAll0.9337
Wafer0.9829
Wine0.8815
WordSynonyms0.7101
Worms0.6169
WormsTwoClass0.7092
Yoga0.8681

Algorithm: