Decomposing one-class support vector machine into an ensemble of one-data support vector machines

2026-06-14 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Advanced, quick

Summary

A new acceleration strategy for the One-Class Support Vector Machine (OCSVM) addresses its scalability issues with large datasets. This approach decomposes the dataset into individual samples, training separate OCSVM models for each single data point. Subsequently, ensemble learning combines these individual models to construct the overall OCSVM model for the entire dataset. Further acceleration is achieved through a data-reduction strategy, which involves training an OCSVM model on the average of the training samples. Experiments, conducted using a Python package, demonstrate that the proposed strategy is faster than traditional OCSVM while maintaining similar classification results. Additionally, this method establishes a one-to-one correspondence between samples and models. Source code is available at https://github.com/ToshiHayashi/ODSVM.

Key takeaway

For Machine Learning Engineers dealing with One-Class Support Vector Machine (OCSVM) scalability issues on large datasets, you should evaluate this decomposition strategy. It offers a significantly faster approach than traditional OCSVM while maintaining similar classification performance. Consider integrating the proposed ensemble and data-reduction techniques into your workflows to improve processing efficiency. You can explore the provided Python package and source code to implement this method.

Key insights

Decomposing OCSVM into an ensemble of one-data models significantly accelerates processing for large datasets.

Principles

• Ensemble learning enhances OCSVM scalability.
• Data decomposition addresses large dataset challenges.
• Averaging training samples enables data reduction.

Method

Decompose the dataset into samples, train individual OCSVM models for each, then combine via ensemble learning. Further accelerate with a data-reduction strategy using an OCSVM on averaged training samples.

In practice

• Utilize the provided Python package.
• Access source code at the GitHub link.
• Leverage sample-model correspondence.

Topics

• One-Class Classification
• Support Vector Machines
• Ensemble Learning
• Scalability Optimization
• Data Reduction

Code references

• ToshiHayashi/ODSVM

Best for: Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• Support Vector Machines (SVMs) - Explained — SVMs classify data by finding the hyperplane that maximizes the margin between classes, defined by support vectors.
• Bootstrapping-based Regularisation for Reducing Individual Prediction Instability in Clinical Risk Prediction Models — A novel regularisation method improves clinical prediction model stability without sacrificing interpretability, crucial for healthcare adoption.
• Universal Transformation of One-Class Classifiers for Unsupervised Anomaly Detection — A dataset folding method converts one-class classifiers into unsupervised anomaly detectors without modifying their core.
• Smart Ensemble Learning Framework for Predicting Groundwater Heavy Metal Pollution — Gaussian copula transformation with stacked ensemble learning robustly predicts groundwater heavy metal pollution.
• Sample Selection Using Multi-Task Autoencoders in Federated Learning with Non-IID Data — Multi-task autoencoders and outlier detection enhance federated learning performance by filtering noisy, redundant, or malicious samples.
• Simplex-Constrained Sparse Bagging: Transitioning from Uniform Priors to Sparse Posteriors in Ensemble Learning — SCSB compresses and calibrates bagging ensembles by optimizing sparse voting weights, improving efficiency and confidence.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.