Learning in the presence of concept recurrence in data stream clustering

Namitha, K.; Kumar, G. Santhosh

doi:10.1186/s40537-020-00354-1

Cited by 9 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This type of drift may be evident and detectable after a while from the start point. ❼ Recurrent or recurring drift refers to the reoccurrence of some foregone distributions or concepts after a while (Namitha and Santhosh Kumar, 2020).…”

Section: Concept Driftmentioning

confidence: 99%

An Experimental Review of the Ensemble-Based Data Stream Classification Algorithms in Non-Stationary Environments

Khezri

Tanha

Samadi

2023

Preprint

View full text Add to dashboard Cite

Section: Concept Driftmentioning

confidence: 99%

An Experimental Review of the Ensemble-Based Data Stream Classification Algorithms in Non-Stationary Environments

Khezri

Tanha

Samadi

2023

Preprint

View full text Add to dashboard Cite

“…e algorithm uses too many parameters that are difficult to tune. Namitha [11] proposed a novel algorithm to identify recurring concepts in data stream clustering. If concept drift is detected, the algorithm retrieves the most matching model from the repository.…”

Section: Related Workmentioning

confidence: 99%

Recurrent Adaptive Classifier Ensemble for Handling Recurring Concept Drifts

Museba

Nelwamondo

Ouahada

et al. 2021

Applied Computational Intelligence and Soft Computing

View full text Add to dashboard Cite

For most real-world data streams, the concept about which data is obtained may shift from time to time, a phenomenon known as concept drift. For most real-world applications such as nonstationary time-series data, concept drift often occurs in a cyclic fashion, and previously seen concepts will reappear, which supports a unique kind of concept drift known as recurring concepts. A cyclically drifting concept exhibits a tendency to return to previously visited states. Existing machine learning algorithms handle recurring concepts by retraining a learning model if concept is detected, leading to the loss of information if the concept was well learned by the learning model, and the concept will recur again in the next learning phase. A common remedy for most machine learning algorithms is to retain and reuse previously learned models, but the process is time-consuming and computationally prohibitive in nonstationary environments to appropriately select any optimal ensemble classifier capable of accurately adapting to recurring concepts. To learn streaming data, fast and accurate machine learning algorithms are needed for time-dependent applications. Most of the existing algorithms designed to handle concept drift do not take into account the presence of recurring concept drift. To accurately and efficiently handle recurring concepts with minimum computational overheads, we propose a novel and evolving ensemble method called Recurrent Adaptive Classifier Ensemble (RACE). The algorithm preserves an archive of previously learned models that are diverse and always trains both new and existing classifiers. The empirical experiments conducted on synthetic and real-world data stream benchmarks show that RACE significantly adapts to recurring concepts more accurately than some state-of-the-art ensemble classifiers based on classifier reuse.

show abstract

“…This concept is in conflict with the goal of detecting evolving cluster structures over time. A recent approach from [12] solves this issue by introducing a two-level architecture with an online component, which clusters the stream data, and an offline component, which saves recurring cluster structures (called concepts) for later use. A drawback stems from the lack of support for arbitrarily shaped clusters, which decreases the usability in applications where no prior information about shape of clusters and data distribution is known.…”

Section: State Of the Art In Data Stream Clustering And Cluster Tracking Applicationsmentioning

confidence: 99%

Unsupervised Detection of State Changes During Operation of Machine Elements

Hillenbrand¹,

Fleischer²

2021

Journal of Machine Engineering

View full text Add to dashboard Cite

Interpretation of sensor data from machine elements is challenging, if no prior knowledge of the system is available. Evaluation methods must adapt surrounding conditions and operation modes. As supervised learning models can be time-consuming to set up, unsupervised learning poses as alternative solution. This paper introduces a new clustering scheme that incorporates iterative cluster retrieval in order to track the clustering results over time. The approach is used to identify changing machine element states such as operating conditions and undesired changes, like incipient damage or wear. We show that knowledge about the evolving clusters can be used to identify operation and failure events. The approach is validated for machine elements with slide and roll contacts, such as ball screws and bearings. The data used has been captured using vibration and acoustic emission sensors. The results show a general applicability to the unsupervised monitoring of machine elements using the proposed approach.

show abstract

Learning in the presence of concept recurrence in data stream clustering

Cited by 9 publications

References 37 publications

An Experimental Review of the Ensemble-Based Data Stream Classification Algorithms in Non-Stationary Environments

An Experimental Review of the Ensemble-Based Data Stream Classification Algorithms in Non-Stationary Environments

Recurrent Adaptive Classifier Ensemble for Handling Recurring Concept Drifts

Unsupervised Detection of State Changes During Operation of Machine Elements

Contact Info

Product

Resources

About