Reliable early classification of time series based on discriminating the classes over time

Mori, Usue; Mendiburu, Alexander; Keogh, Eamonn; Lozano, José A.

doi:10.1007/s10618-016-0462-1

Cited by 71 publications

(91 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note that, for each database, the accuracy evolves differently as more data becomes available. As shown in [14], in some databases, the accuracy is strictly increasing. In others, in contrast, the accuracy becomes stable after some point, and in some databases, the accuracy even drops after some point due to the noise that the additional data incorporates.…”

Section: Prediction Stepmentioning

confidence: 88%

“…Also, Ghalwash et al [11] mention early stock crisis identification; Bregón et al [12] apply early classification to classify different types of faults in a simulated industrial plant; Hatami and Chira [13] attempt to classify a set of different odors as early as possible by using odor signals obtained from a set of sensors with the aim of identifying chemical leaks. Finally, in Mori et al [14], an early classification approach is applied to detect and identify bird songs as early as possible, with the objective of saving memory and battery life of the recording devices.…”

Section: Introductionmentioning

confidence: 99%

“…This is exactly the aim of early classification, finding a trade-off between two conflicting objectives: the accuracy of the predictions and their earliness [9]. However, the degree of conflict between these two objectives changes drastically from database to database and, as more data becomes available, the accuracy evolves differently in each database [14]. So, the problem of early classification can be described as an optimization problem in which two conflicting objectives must be optimized at the same time.…”

Section: Introductionmentioning

confidence: 99%

“…. , (T S n , CL n )}, the aim is to build a model that is able to predict the class label of new unlabeled time series as accurately as possible [1,2,9,14].…”

mentioning

confidence: 99%

“…A new method, denominated Early Classification of Time Series based on Discriminating Classes Over Time (ECDIRE) [14], lies in between these two methods. The main objective of this method is to analyze the evolution of the accuracy of a set of probabilistic classifiers over time, with the aim of identifying the timestamps from whence it is safe to make predictions.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness

Mori

Mendiburu

Dasgupta

et al. 2018

IEEE Trans. Neural Netw. Learning Syst.

Self Cite

147

View full text Add to dashboard Cite

The problem of early classification of time series appears naturally in contexts where the data, of temporal nature, is collected over time, and early class predictions are interesting or even required. The objective is to classify the incoming sequence as soon as possible, while maintaining suitable levels of accuracy in the predictions. Thus, we can say that the problem of early classification consists in optimizing two objectives simultaneously: accuracy and earliness. In this context, we present a method for early classification of time series based on combining a set of probabilistic classifiers together with a stopping rule. This stopping rule will act as a trigger and will tell us when to output a prediction or when to wait for more data, and it's main novelty lies in the fact that it is built by explicitly optimizing a cost function based on accuracy and earliness. We have selected a large set of benchmark datasets and 4 other state-ofthe-art early classification methods and we have evaluated and compared our framework obtaining superior results in terms of both earliness and accuracy.

show abstract

Section: Prediction Stepmentioning

confidence: 88%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…. , (T S n , CL n )}, the aim is to build a model that is able to predict the class label of new unlabeled time series as accurately as possible [1,2,9,14].…”

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness

Mori

Mendiburu

Dasgupta

et al. 2018

IEEE Trans. Neural Netw. Learning Syst.

Self Cite

147

View full text Add to dashboard Cite

show abstract

A novel approach for early malware detection

Sharma

Singh

2020

Trans Emerging Tel Tech

View full text Add to dashboard Cite

Early classification of time series is valuable in many real‐world applications such as early disease prediction, early disaster prediction, and patient monitoring where data are generated over time. The main objective of early classification is to provide a reliable class prediction earliest in time. In general, whenever the early prediction time improves, the prediction accuracy decreases. Thus, the trade‐off between earliness and accuracy needs to be addressed. In this article, we proposed an optimization‐based early classification model for time series data using early stopping rules (ESRs) and a series of probabilistic classifiers. ESRs are developed through particle swarm optimization by minimizing the well‐defined cost function that considers the missclassification cost and delaying decision cost simultaneously. The experimental results on 30 standard datasets demonstrate good performance for early classification in comparison to state of the art methods. Also, the proposed model is tested for early malware detection on a real dataset and shows decent performance by balancing the accuracy and earliness.

show abstract

Early prediction of Ibex 35 movements

Miranda‐García

Vargas

Mori

et al. 2022

Journal of Forecasting

Self Cite

View full text Add to dashboard Cite

In this paper, we examine the early predictability of the market's directional movement using intraday high‐frequency data (695,764 observations) from an stock index (Ibex 35 Index) to provide, either private or institutional investors, an early warning system based on an “early indicator” of the financial market fluctuations with an optimal combination of the two more relevant variables for this strategy, accuracy, and earliness. A novel supervised machine learning early classification technique (Artificial Intelligence) has been applied, for the first time, to the high‐frequency time series of both price and certain technical indicators. The results obtained allow us to assert that the intraday movement of the Ibex 35 can be predicted with acceptable levels of accuracy 24 min after the start of the session and to establish certain informative intraday hourly patterns. Consequently, different indicators of precision and earliness in the session are generated, obtaining that, after a certain point in the session, no gains in precision are generated.

show abstract

Reliable early classification of time series based on discriminating the classes over time

Cited by 71 publications

References 22 publications

Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness

Early Classification of Time Series by Simultaneously Optimizing the Accuracy and Earliness

A novel approach for early malware detection

Early prediction of Ibex 35 movements

Contact Info

Product

Resources

About