A framework for on-demand classification of evolving data streams

Aggarwal, Charų C.; Wang, Jianyong; Yu, Philip S.

doi:10.1109/tkde.2006.69

Cited by 106 publications

(65 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address feature evolution, they proposed feature set homogenization technique. Aggarwal et al proposed a model which is able to adapt with changes in data streams(concept evolution) [22]. They proposed On Demand Classification which is able to dynamically determine appropriate window size for past training data.…”

Section: Incremental Learning and Ensemble Methodsmentioning

confidence: 99%

A review on data stream classification approaches

Homayoun

Ahmadzadeh

2016

JACST

View full text Add to dashboard Cite

Stream data is usually in vast volume, changing dynamically, possibly infinite, and containing multi-dimensional features. The attention towards data stream mining is increasing as regards to its presence in wide range of real-world applications, such as e-commerce, banking, sensor data and telecommunication records. Similar to data mining, data stream mining includes classification, clustering, frequent pattern mining etc. techniques; the special focus of this paper is on classification methods invented to handle data streams. Early methods of data stream classification needed all instances to be labeled for creating classifier models, but there are some methods (SemiSupervised Learning and Active Learning) in which unlabeled data is employed as well as labeled data. In this paper, by focusing on ensemble methods, semi-supervised and active learning, a review on some state of the art researches is given.

show abstract

Section: Incremental Learning and Ensemble Methodsmentioning

confidence: 99%

A review on data stream classification approaches

Homayoun

Ahmadzadeh

2016

JACST

View full text Add to dashboard Cite

show abstract

“…These approaches are basically of two types, namely single model incremental approach and hybridbatch incremental approach. In single model incremental approach ( proposed in [1]) use the concept of a single model. This model is continuously updated with the new features or data appearing in the stream data.…”

Section: IImentioning

confidence: 99%

A Proposed Approach to Classification and Novel Class Detection in Data Streams

Singh¹,

Chandak²

2015

International Journal of Advanced Research in Computer and Comm

View full text Add to dashboard Cite

Abstract:In the field of data mining, data streams play an important role. In order to make intelligent use of data streams, first we need to handle them properly. The notation data stream itself represents the nature of such data. They possess the property of being dynamic and continuous in nature. That is the data keeps on changing the features and properties with time. Due to the above mentioned properties of data streams, various challenges are posed by them to researchers. These challenges are infinite-length, concept-evolution, feature-evolution and concept-drift. Infinite-length is due to the continuous nature of data. Concept-evolution is due to the new emerging classes. Concept-drift is due to the drifting concept of the stream and feature-evolution is there because of the changing features. These challenges have been well studied by various researchers and they have proposed various approaches to handle them. In this paper we also try to propose a strategy based on strings to handle the problems of infinite-length, concept-evolution and concept-drift.

show abstract

“…Most data stream classification techniques handle conceptdrift using a number of different techniques [1], [3]- [6], [11]- [17]. Two popular alternatives to handle the massive volume of data streams and concept-drift issues are the single-model incremental approach, and hybrid batchincremental approach.…”

Section: Related Workmentioning

confidence: 99%

“…In the single model approach, a single model is dynamically maintained with the new data. For example, [5] incrementally updates a decision tree with incoming data, and [1] incrementally updates micro-clusters in the model with the new data. The batch-incremental approach builds each classification model using a batch learning technique.…”

Section: Related Workmentioning

confidence: 99%

“…One pass learning is required because of the massive volume and continuous delivery of the data stream. Concept-drift occurs in dynamic streams, and many different techniques have been designed with the same goal of maintaining an upto-date classification model [1]- [3], [5], [6], [11]. Another issue of considerable recent interest is that of conceptevolution, which refers to the emergence of a new class.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Stream Classification with Recurring and Novel Class Detection Using Class-Based Ensemble

Al-Khateeb

Masud

Khan

et al. 2012

2012 IEEE 12th International Conference on Data Mining

View full text Add to dashboard Cite

Abstract-Concept-evolution has recently received a lot of attention in the context of mining data streams. Conceptevolution occurs when a new class evolves in the stream. Although many recent studies address this issue, most of them do not consider the scenario of recurring classes in the stream. A class is called recurring if it appears in the stream, disappears for a while, and then reappears again. Existing data stream classification techniques either misclassify the recurring class instances as another class, or falsely identify the recurring classes as novel. This increases the prediction error of the classifiers, and in some cases causes unnecessary waste in memory and computational resources. In this paper we address the recurring class issue by proposing a novel "class-based" ensemble technique, which substitutes the traditional "chunk-based" ensemble approaches and correctly distinguishes between a recurring class and a novel one. We analytically and experimentally confirm the superiority of our method over state-of-the-art techniques.

show abstract

A framework for on-demand classification of evolving data streams

Cited by 106 publications

References 24 publications

A review on data stream classification approaches

A review on data stream classification approaches

A Proposed Approach to Classification and Novel Class Detection in Data Streams

Stream Classification with Recurring and Novel Class Detection Using Class-Based Ensemble

Contact Info

Product

Resources

About