On-Line Dynamic Time Warping for Streaming Time Series

Oregi, Izaskun; Pérez, Aritz; Ser, Javier Del; Lozano, José A.

doi:10.1007/978-3-319-71246-8_36

Cited by 17 publications

(17 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Then we will study the effect of the forgetting mechanism by analyzing the results drawn by a 1-NN classifier over transitions between stationary stream intervals. Built upon preliminary findings reported in (Oregi et al, 2017b), this work extends this previous work by proposing a complete framework for computing on-line similarity measures that considers any kind of inner cost functions and similarity-based on-line learning scenarios. We describe in depth the concepts behind the design of the proposed OESM, and discuss its performance under different configurations over a much broader set of experiments.…”

Section: Introductionmentioning

confidence: 78%

On-line Elastic Similarity Measures for time series

Oregi

Pérez

Ser

et al. 2019

Pattern Recognition

Self Cite

View full text Add to dashboard Cite

The way similarity is measured among time series is of paramount importance in many data mining and machine learning tasks. For instance, Elastic Similarity Measures are widely used to determine whether two time series are similar to each other. Indeed, in off-line time series mining, these measures have been shown to be very effective due to their ability to handle time distortions and mitigate their effect on the resulting distance. In the on-line setting, where available data increase continuously over time and not necessary in a stationary manner, stream mining approaches are required to be fast with limited memory consumption and capable of adapting to different stationary intervals. In this sense, the computational complexity of Elastic Similarity Measures and their lack of flexibility to accommodate different stationary intervals, make these similarity measures incompatible with the requirements mentioned. To overcome these issues, this paper adapts the family of Elastic Similarity Measures-which includes Dynamic Time Warping, Edit Distance, Edit Distance for Real Sequences and Edit Distance with Real Penalty-to the on-line setting. The proposed adaptation is based on two main ideas: a forgetting mechanism and the incremental computation. The former makes the similarity consistent with streaming time series characteristics by giving more importance to recent observations, whereas the latter reduces the computational complexity by avoiding unnecessary computations. In order to assess the behavior of the proposed similarity measure in on-line settings, two different experiments have been carried out. The first aims at showing the efficiency of the proposed adaptation, to do so we calculate and compare the computation time for the elastic measures and their on-line adaptation. By analyzing the results drawn from a distance-based streaming machine learning model, the second experiment intends to show the effect of the forgetting mechanism on the resulting similarity value. The experimentation shows, for the aforementioned Elastic Similarity Measures, that the proposed adaptation meets the memory, computational complexity and flexibility constraints imposed by streaming data.

show abstract

Section: Introductionmentioning

confidence: 78%

On-line Elastic Similarity Measures for time series

Oregi

Pérez

Ser

et al. 2019

Pattern Recognition

Self Cite

View full text Add to dashboard Cite

show abstract

“…The fact that DTW can only be used to match sequences with same shape, that is, content means that it cannot be used for online collision detection, as it would only be able to detect collision after the entire sequence has been recorded. Solutions proposed and described in 19 offer increase in speed and a modified measure of dissimilarity of compared signals. However, they are still not suitable for application in collision detection, as they presume the same starting and ending point of compared signals.…”

Section: Modified Dynamic Time Warpingmentioning

confidence: 99%

“…Compared to traditional DTW, 18 it offers two advantages of key importance for Collision detection application. Most importantly, unlike, [18][19][20] it allows matching signals with different contents. That means one signal can be compared to a part of another signal with similar content or shape.…”

Section: Introductionmentioning

confidence: 99%

“…The fourth section explains DTW method, as one of the most efficient ways for matching two time sequences. 18,19 Additionally, it presents the modification of DTW which was developed for the purpose of the collision detection algorithm itself. Modification brings the benefit of being able to match subsequence to the corresponding section of the complete sequence, making the modified DTW applicable in online collision detection, rather than only in offline analysis.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Collision Detection on Industrial Robots in Repetitive Tasks Using Modified Dynamic Time Warping

Gordić¹,

Jovanović²

2019

Robotica

View full text Add to dashboard Cite

SUMMARYThis paper presents a non-model-based collision detection algorithm for robots without external sensors and with closed control architecture. A reference signal of repetitive motion is recorded from the robot operation. To detect collisions, the reference is compared with measurements from the robot. One of the key contributions is a novel approach to optimal matching of compared signals, which is ensured by the newly developed modified Dynamic Time Warping (mDTW) method presented in this paper. One of the main improvements of the mDTW is that it enables comparing a signal with the most similar section of the other signal. Partial matching also enables online application of time warping principles and reduces the time and computation resources needed to perform matching. In addition to mDTW, two complementary decision rules are developed to identify collisions. The first rule, based on the absolute difference between compared matched samples, uses statistically determined thresholds to perform rapid detection of unambiguous collisions. The second rule is based on Eigen values of the covariance matrix of matched samples, and it employs its higher sensitivity to detect collisions with lower intensity. Results from experimental validation of the proposed collision algorithm on two industrial robots are shown.

show abstract

“…DTW is memory greedy on trace size, a similar problem arises when dealing with streaming traces. Oregi et al [20] and Martins et al [15] present a generalization of DTW for large streaming data. They propose the use of incremental computation of the cost matrix complemented with a weighted event distance function adding event positions.…”

Section: Related Workmentioning

confidence: 99%

Scalable comparison of JavaScript V8 bytecode traces

Arteaga

Monperrus

Baudry

2019

Proceedings of the 11th ACM SIGPLAN International Workshop on Virtual Machines and Intermediate Languages

View full text Add to dashboard Cite

The comparison and alignment of runtime traces are essential, e.g., for semantic analysis or debugging. However, naive sequence alignment algorithms cannot address the needs of the modern web: (i) the bytecode generation process of V8 is not deterministic; (ii) bytecode traces are large.We present STRAC, a scalable and extensible tool tailored to compare bytecode traces generated by the V8 JavaScript engine. Given two V8 bytecode traces and a distance function between trace events, STRAC computes and provides the best alignment. The key insight is to split access between memory and disk. STRAC can identify semantically equivalent web pages and is capable of processing huge V8 bytecode traces whose order of magnitude matches today's web like https: //2019.splashcon.org, which generates approx. 150k of V8 bytecode instructions.

show abstract

On-Line Dynamic Time Warping for Streaming Time Series

Cited by 17 publications

References 25 publications

On-line Elastic Similarity Measures for time series

On-line Elastic Similarity Measures for time series

Collision Detection on Industrial Robots in Repetitive Tasks Using Modified Dynamic Time Warping

Scalable comparison of JavaScript V8 bytecode traces

Contact Info

Product

Resources

About