## Efficient Retrieval of Similar Time Sequences Under Time Warping (1997)

### Download From

IEEE### Download Links

- [www.cs.cmu.edu]
- [www.cs.cmu.edu]
- [www.cs.cmu.edu]
- DBLP

### Other Repositories/Bibliography

Citations: | 174 - 3 self |

### BibTeX

@INPROCEEDINGS{Yi97efficientretrieval,

author = {Byoung-kee Yi and H. V. Jagadish and Christos Faloutsos},

title = {Efficient Retrieval of Similar Time Sequences Under Time Warping},

booktitle = {},

year = {1997},

pages = {201--208}

}

### Years of Citing Articles

### OpenURL

### Abstract

Fast similarity searching in large time-sequence databases has attracted a lot of research interest [1, 5, 2, 6, 3, 10]. All of them use the Euclidean distance (L 2 ), or some variation of L p metrics. L p metrics lead to efficient indexing, thanks to feature extraction (e.g., by keeping the first few DFT coefficients) and subsequent use of fast spatial access methods for the points in feature space. In this work we examine a popular, field-tested dissimilarity function, the "time warping" distance function which permits local accelerations and decelerations in the rate of the signals or sequences. This function is natural and suitable for several applications, like matching of voice, audio and medical signals (e.g., electrocardiograms) However, from the indexing viewpoint it presents two major challenges: (a) it does not lead to any natural "features", precluding the use of spatial access methods (b) it is quadratic (O(len 1 len 2 )) on the length of the sequences involved. Here we ...

### Citations

3119 |
Introduction to Modern Information Retrieval
- Salton
- 1983
(Show Context)
Citation Context ...fl FM+LB: The proposed "combined" method, which enables both filters in Figure 3. To measure how many false-dismissals are introduced by FastMap, we use the "recall" concept from I=-=nformation Retrieval[11]-=-. Definition 3 Recall is defined as follows: recall j retrieved and relevant relevant The (ideal) recall value of 1.0 means there are no false-dismissals, while a recall value of 0.0 means that no rel... |

1487 |
Fundamentals of Speech Recognition
- Rabiner, Juang
- 1993
(Show Context)
Citation Context ...; : : : ; x n i. Then, we want to allow the stuttering transformation on a sequence, possibly with a penalty ("cost"): ffl stutter i (~x) : repeats x i and shifts the elements to the right. =-=Following [9]-=-, for any non-null sequences ~x and ~y, the (dynamic) time warping distance is defined as follows. Symbol Definition D p L p -based distance function D base base distance function, e.g., D 1 or D 2 Dw... |

414 | Faloutsos,â€œEfficient similarity search in sequence databases
- Agrawal, Christos
- 1993
(Show Context)
Citation Context ...nces Under Time Warping Byoung-Kee Yi H.V. Jagadish y Christos Faloutsos z October 9, 1997 Abstract Fast similarity searching in large time-sequence databases has attracted a lot of research interest =-=[1, 5, 2, 6, 3, 10]-=-. All of them use the Euclidean distance (L 2 ), or some variation of L p metrics. L p metrics lead to efficient indexing, thanks to feature extraction (e.g., by keeping the first few DFT coefficients... |

409 | FastMap: A Fast Algorithm for Indexing, Data-Mining and
- Faloutsos, Lin
- 1995
(Show Context)
Citation Context ...e. We would like to have a very fast matching technique, and ideally even an indexing technique for this purpose. In this paper we propose two such techniques. The first technique is based on FastMap =-=[4]-=-. The idea here is to make use of the given distance measures to map sequences into points in k-d space, and to finally build an index structure. The other technique we propose defines a new distance ... |

382 |
Time Warps, String Edits, and Macromolecules: The Theory and Practice of Sequence Comparison
- Sankoff, Kruskal
- 1983
(Show Context)
Citation Context ...tering as needed at no cost. Defined as a recurrence, the time-warping distance can be computed by a dynamic programming algorithm (see Appendix A or [9]) whose complexity is O(j~xj \Theta j~yj). See =-=[12]-=- for more details and other variants of the basic algorithm. Figure 1 shows two time sequences, before and after the time warping. The sequences are mixtures of similar harmonics: x(t) = 10 sin(0:5t) ... |

198 | Fast similarity search in the presence of noise, scaling, and translation in timeseries databases
- Agrawal, Lin, et al.
- 1995
(Show Context)
Citation Context ...nces Under Time Warping Byoung-Kee Yi H.V. Jagadish y Christos Faloutsos z October 9, 1997 Abstract Fast similarity searching in large time-sequence databases has attracted a lot of research interest =-=[1, 5, 2, 6, 3, 10]-=-. All of them use the Euclidean distance (L 2 ), or some variation of L p metrics. L p metrics lead to efficient indexing, thanks to feature extraction (e.g., by keeping the first few DFT coefficients... |

137 | On sirnilarity-based queries for time series data
- Rafiei
- 1999
(Show Context)
Citation Context ...nces Under Time Warping Byoung-Kee Yi H.V. Jagadish y Christos Faloutsos z October 9, 1997 Abstract Fast similarity searching in large time-sequence databases has attracted a lot of research interest =-=[1, 5, 2, 6, 3, 10]-=-. All of them use the Euclidean distance (L 2 ), or some variation of L p metrics. L p metrics lead to efficient indexing, thanks to feature extraction (e.g., by keeping the first few DFT coefficients... |

99 | On similarity queries for time-series data: constraint specification and implementation
- Goldin, Kanellakis
- 1995
(Show Context)
Citation Context ...tock price time sequences were generated by extracting 150 most recent(as of 6/5/96) daily high values from 640 stocks. These time sequences were normalized by subtracting the average, as was done in =-=[6, 2]. -8 -6 -4 -2 0 2 4 -=-6 8 0 20 40 60 80 100 120 140 Time "Synthetic" -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 100 200 300 400 500 600 700 Amplitude Time "ECG" -5 -4 -3 -2 -1 0 1 2 3 0 20 40 60 80 100 120 140 160... |

98 | Fast nearest neighbor search in medical image databases
- Korn, Sidiropoulos, et al.
- 1996
(Show Context)
Citation Context ...ffl. As we have shown in a previous paper[5], lower-bounding the actual distance with another distance is a condition that guarantees no false dismissals for range queries and nearest neighbor queries=-=[8]-=-. Algorithm 2 describes how range queries can be processed. Other types of queries can be handled similarly. In the filtering step, irrelevant sequences are filtered out quickly because the D lb dista... |

67 |
Yannis Manolopoulos. Fast subsequence matching in time-series databases
- Faloutsos, Ranganathan
(Show Context)
Citation Context |

66 | Similarity-Based Queries
- Jagadish, Mendelzon, et al.
- 1995
(Show Context)
Citation Context ...es. Rafiei and Mendelzon [10] extend previous work by proposing techniques to handle moving average and time scaling(i.e., globally stretching or shrinking of the time axis), but not time warping. In =-=[7]-=-, we et al develop domain independent framework for defining queries in terms of similarity of objects. Our framework has three components: a pattern language, a transformation rule language, and a qu... |

29 | A signature technique for similarity-based queries
- Faloutsos, Jagadish, et al.
- 1997
(Show Context)
Citation Context |

12 |
The input-state space approach to the prediction of auroral geomagnetic activity from solar wind variables
- Vassiliadis
- 1993
(Show Context)
Citation Context ...oft"); digital audio/voice clips ("find clips that sound like a given person") [1, 5]; scientific databases ("find times in the past that had similar solar magnetic wind patterns w=-=ith the ones today" [14]). In the -=-area of speech recognition, this problem has been studied extensively, and is called the "(dynamic) time warping". Virtually all speech recognition systems speed-up and slow down portions of... |

1 |
Raghu Ramakrishnan. Sequnce query processing
- Seshadri, Livny
- 1994
(Show Context)
Citation Context ... query language. The framework can be tuned to the needs of a specific application domain, such as time sequences, molecules, text strings or images, by the choice of these languages. Sheshadri et al =-=[13]-=- suggest a new data model and an algebraic language for sequences in general. They also propose a sophisticated optimization technique, but do not mention about similarity among sequences and query pr... |