![](bg8.png)
not work well with respect to the Asiac data set (possi-
bly because of its size in terms of number of records).
In general, the proposed DSSM and MPSSM mecha-
nisms, performed better or as well as Standard DTW
or Enhanced Sub-sequence based DTW, but with an
improved run time (much improved in some cases).
From the results presented in Table 6 an argument can
be made that DSSM produced the best results.
7 CONCLUSION
In this paper, two DTW mechanisms have been pro-
posed founded on the concept of motifs: (i) the
Differential Sub-Sequence Motifs (DSSM) mecha-
nism and (ii) the Matrix Profile Sub-Sequence Motifs
(MPSSM) mechanism. Both were directed at speed-
ing up the DTW process without adversely affecting
accuracy. The operation of the proposed mechanisms
was compared with Standard DTW and Enhanced
Sub-sequence based DTW, using a kNN classification
model with k = 1 and ten time series data sets of a va-
riety of sizes; taken from the UEA and UCR (Univer-
sity of East Anglia and University of California River-
side) Time Series Classification Repository (Bagnall
et al., 2017). The evaluation demonstrated that the
proposed mechanisms outperformed the comparator
mechanisms in nine out of ten cases with respect to
run time without adversely affecting classification ac-
curacy. Out of the two proposed mechanisms DSSM
gave the best performance.
REFERENCES
Alshehri, M., Coenen, F., and Dures, K. (2019a). Effec-
tive sub-sequence-based dynamic time warping. In In-
ternational Conference on Innovative Techniques and
Applications of Artificial Intelligence, pages 293–305.
Springer.
Alshehri, M., Coenen, F., and Dures, K. (2019b). Sub-
sequence-based dynamic time warping. In KDIR,
pages 274–281.
Bagnall, A., Lines, J., Bostrom, A., Large, J., and Keogh,
E. (2017). The great time series classification bake
off: a review and experimental evaluation of recent
algorithmic advances. Data Mining and Knowledge
Discovery, 31(3):606–660.
Brunello, A., Marzano, E., Montanari, A., and Sciavicco,
G. (2018). A novel decision tree approach for the
handling of time series. In International Conference
on Mining Intelligence and Knowledge Exploration,
pages 351–368. Springer.
Ebadati, O. and Mortazavi, M. (2018). An efficient hybrid
machine learning method for time series stock market
forecasting. Neural Network World, 28(1):41–55.
Fawaz, H. I., Forestier, G., Weber, J., Idoumghar, L., and
Muller, P.-A. (2019). Deep learning for time series
classification: a review. Data Mining and Knowledge
Discovery, 33(4):917–963.
Gamboa, J. C. B. (2017). Deep learning for time-series
analysis. arXiv preprint arXiv:1701.01887.
Karevan, Z. and Suykens, J. A. (2020). Transductive lstm
for time-series prediction: An application to weather
forecasting. Neural Networks, 125:1–9.
Niennattrakul, V. and Ratanamahatana, C. A. (2009).
Learning dtw global constraint for time series classifi-
cation. arXiv preprint arXiv:0903.0041.
Phinyomark, A. and Scheme, E. (2018). An investigation
of temporally inspired time domain features for elec-
tromyographic pattern recognition. In 2018 40th An-
nual International Conference of the IEEE Engineer-
ing in Medicine and Biology Society (EMBC), pages
5236–5240. IEEE.
Rakthanmanon, T., Campana, B., Mueen, A., Batista, G.,
Westover, B., Zhu, Q., Zakaria, J., and Keogh, E.
(2012). Searching and mining trillions of time se-
ries subsequences under dynamic time warping. In
Proceedings of the 18th ACM SIGKDD international
conference on Knowledge discovery and data mining,
pages 262–270. ACM.
Sakoe, H. and Chiba, S. (1978). Dynamic programming
algorithm optimization for spoken word recognition.
IEEE transactions on acoustics, speech, and signal
processing, 26(1):43–49.
Silva, D. F., Giusti, R., Keogh, E., and Batista, G. E. (2018).
Speeding up similarity search under dynamic time
warping by pruning unpromising alignments. Data
Mining and Knowledge Discovery, pages 1–29.
Torkamani, S. and Lohweg, V. (2017). Survey on time se-
ries motif discovery. Wiley Interdisciplinary Reviews:
Data Mining and Knowledge Discovery, 7(2):e1199.
Yeh, C.-C. M., Zhu, Y., Ulanova, L., Begum, N., Ding,
Y., Dau, H. A., Silva, D. F., Mueen, A., and Keogh,
E. (2016). Matrix profile i: all pairs similarity joins
for time series: a unifying view that includes mo-
tifs, discords and shapelets. In 2016 IEEE 16th in-
ternational conference on data mining (ICDM), pages
1317–1322. Ieee.
Motif-based Classification using Enhanced Sub-Sequence-Based Dynamic Time Warping
191