HOT SAX: efficiently finding the most unusual time series subsequence

By Eamonn J. Keogh, Jessica Lin, Ada Wai-Chee Fu. Industrial Conference on Data Mining, 2005

850 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Economics and Econometrics, Signal Processing

View Paper: https://www.semanticscholar.org/paper/ee0c3888c8e918ee669a9efc59094add6d72a68f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Economics and Econometrics

In Signal Processing

Citations in the corpus

A review on time series data mining (2011, methodology): “Wei et al. (2006) adopt the SAX representation for anomaly detection on the basis of shape.”
Hierarchical Probabilistic Models for Group Anomaly Detection (2011, background): “To do this, we need to define a set of features for the groups (Chan and Mahoney, 2005; Keogh et al., 2005).”
Anomaly Detection for Discrete Sequences: A Survey (2012, methodology): “Several faster variations have been proposed that can run in approximately linear time in the context of continuous time series [67], [68], [72], [73], [74], and can be extended to discrete sequences.”
Trajectory-Based Anomalous Event Detection (2008, methodology): “To perform this test, we compared the proposed method with another simple yet very effective outlier detection technique, based on the concept of discords [39], [40]: a discord is defined as the trajectory maximizing”
Using dynamic time warping distances as features for improved time series classification (2016, methodology): “Besides classification, SAX has been also employed for time series clustering and anomaly detection (Keogh et al. 2005; Lin et al. 2012).”
Group Anomaly Detection using Flexible Genre Models (2011, methodology): “To handle distribution-based anomalies, we can design features for groups and then treat them as points [5, 6].”
Cyberspace Security Using Adversarial Learning and Conformal Prediction (2015, background): “, DWT, DFT) in order to find discords and most unusual time series subsequences [63].”
Why we search: visualizing and predicting user behavior (2007, methodology): “We believe that even if different alignment techniques [8][9] are used, this visualization is still informative.”
Experiencing SAX: a novel symbolic representation of time series (2007, methodology): “We have since then proposed more algorithms on anomaly detection (HOT SAX) (Keogh et al. 2005), contrast sets mining (Lin and Keogh 2006), visualization by time series bitmaps (Kumar et al.”
Spatio-Temporal Data Mining (2017, methodology): “This includes approaches for univariate time series [Bishop 1994; Keogh et al. 2005, 2002; Wei et al. 2005] as well as multivariate time series [Baragona and Battaglia 2007; Chen et al. 2008; Galeano et al. 2006; Takeishi and Yairi 2014] A survey of anomaly detection approaches for temporal data is…”
LSTM-based Encoder-Decoder for Multi-sensor Anomaly Detection (2016, background): “The ﬁrst three are taken from (Keogh et al., 2005) whereas the engine dataset is a proprietary one encountered in a real-life project.”
A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection (2023, neutral): “Traditionally,distance-based[137]anddistributionaltech-niques[138]havebeenwidelyusedfordetectingirregular-itiesintimeseriesdata.”
Large-Scale Unusual Time Series Detection (2015, methodology): “We propose using Principal Component Analysis (PCA) together with multi-dimensional anomaly detection to identify unusual time series in a large collections of time series.”