Online Learning With Adaptive Rebalancing in Nonstationary Environments

Malialis, Kleanthis; Panayiotou, Christos G.

doi:10.1109/tnnls.2020.3017863

Cited by 28 publications

(4 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Adaptive REBAlancing (AREBA) [6], [28] effectively addresses the dual-problem of concept drift and class imbalance. It uses a passive approach (memory-based) to handle drift, and a dynamic rebalancing mechanism to handle imbalance.…”

Section: B Class Imbalancementioning

confidence: 99%

“…Without this mechanism, the initial class imbalance problem would still persist in the memory (queue)based system. This is stated in Line 42 and the details are provided in the original paper [6].…”

Section: B Passive Approachmentioning

confidence: 99%

“…Contrary to the traditional framework where active and passive approaches have been applied in isolation, we propose a method which combines the two in synergy to take advantage of both. • We propose the Hybrid-Adaptive REBAlancing (HAREBA) method, which significantly extends AREBA [6]. The original method is passive, i.e., it implicitly handles drift by using memory and incremental learning; it also uses a dynamic mechanism to constantly maintain balance within the memory.…”

mentioning

confidence: 99%

See 2 more Smart Citations

A Hybrid Active-Passive Approach to Imbalanced Nonstationary Data Stream Classification

Malialis¹,

Roveri²,

Alippi³

et al. 2022

Preprint

View full text Add to dashboard Cite

In real-world applications, the process generating the data might suffer from nonstationary effects (e.g., due to seasonality, faults affecting sensors or actuators, and changes in the users' behaviour). These changes, often called concept drift, might induce severe (potentially catastrophic) impacts on trained learning models that become obsolete over time, and inadequate to solve the task at hand. Learning in presence of concept drift aims at designing machine and deep learning models that are able to track and adapt to concept drift. Typically, techniques to handle concept drift are either active or passive, and traditionally, these have been considered to be mutually exclusive. Active techniques use an explicit drift detection mechanism, and re-train the learning algorithm when concept drift is detected. Passive techniques use an implicit method to deal with drift, and continually update the model using incremental learning. Differently from what present in the literature, we propose a hybrid alternative which merges the two approaches, hence, leveraging on their advantages. The proposed method called Hybrid-Adaptive REBAlancing (HAREBA) significantly outperforms strong baselines and state-of-the-art methods in terms of learning quality and speed; we experiment how it is effective under severe class imbalance levels too.

show abstract

Section: B Class Imbalancementioning

confidence: 99%

Section: B Passive Approachmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

A Hybrid Active-Passive Approach to Imbalanced Nonstationary Data Stream Classification

Malialis¹,

Roveri²,

Alippi³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…3. We provide new insights into learning from nonstationary and imbalanced data streams, which constitutes a challenging and largely unexplored area even in the presence of supervision [2,3,10].…”

Section: Introductionmentioning

confidence: 99%