# Data-efficient Online Classification with Siamese Networks and Active Learning

arXiv: https://arxiv.org/abs/2010.01659 · HTML: https://mlsanity.com/paper/2010.01659/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Oct 2020.

## Built on by (most cited first, 7 of 7)

- [Nonstationary data stream classification with online active learning and siamese neural networks](https://mlsanity.com/paper/2210.01090/) (2022)
- [Autoencoder-based Anomaly Detection in Streaming Data with Incremental Learning and Concept Drift Adaptation](https://mlsanity.com/paper/2305.08977/) (2023)
- [A Hybrid Active-Passive Approach to Imbalanced Nonstationary Data Stream Classification](https://mlsanity.com/paper/2210.04949/) (2022)
- [Data augmentation on-the-fly and active learning in data stream classification](https://mlsanity.com/paper/2210.06873/) (2022)
- [Similarity Learning based Few Shot Learning for ECG Time Series Classification](https://mlsanity.com/paper/2202.00612/) (2022)
- [Benchmarking of Query Strategies: Towards Future Deep Active Learning](https://mlsanity.com/paper/2312.05751/) (2023)
- [SiameseDuo++: Active Learning from Data Streams with Dual Augmented Siamese Networks](https://mlsanity.com/paper/2504.04613/) (2025)

## Abstract (arXiv)

An ever increasing volume of data is nowadays becoming available in a streaming manner in many application areas, such as, in critical infrastructure systems, finance and banking, security and crime and web analytics. To meet this new demand, predictive models need to be built online where learning occurs on-the-fly. Online learning poses important challenges that affect the deployment of online classification systems to real-life problems. In this paper we investigate learning from limited labelled, nonstationary and imbalanced data in online classification. We propose a learning method that synergistically combines siamese neural networks and active learning. The proposed method uses a multi-sliding window approach to store data, and maintains separate and balanced queues for each class. Our study shows that the proposed method is robust to data nonstationarity and imbalance, and significantly outperforms baselines and state-of-the-art algorithms in terms of both learning speed and performance. Importantly, it is effective even when only 1% of the labels of the arriving instances are available.