![[2602.12391] High-dimensional Level Set Estimation with Trust Regions and Double Acquisition Functions](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12391] High-dimensional Level Set Estimation with Trust Regions and Double Acquisition Functions
Summary
This paper presents TRLSE, a novel algorithm for high-dimensional level set estimation, enhancing sample efficiency through dual acquisition functions in active learning settings.
Why It Matters
High-dimensional level set estimation is crucial for various applications, including machine learning and data analysis. The proposed TRLSE algorithm addresses the challenges of limited initial data and exponential search volumes, offering a more efficient approach to classifier construction. This research contributes to the advancement of active learning techniques, making it relevant for practitioners and researchers in machine learning.
Key Takeaways
- TRLSE improves sample efficiency in high-dimensional level set estimation.
- The algorithm utilizes dual acquisition functions for better accuracy.
- Theoretical analysis supports TRLSE's effectiveness over existing methods.
- Extensive evaluations demonstrate TRLSE's performance on synthetic and real-world problems.
- Active learning techniques are enhanced, benefiting various applications.
Computer Science > Machine Learning arXiv:2602.12391 (cs) [Submitted on 12 Feb 2026] Title:High-dimensional Level Set Estimation with Trust Regions and Double Acquisition Functions Authors:Giang Ngo, Dat Phan Trong, Dang Nguyen, Sunil Gupta View a PDF of the paper titled High-dimensional Level Set Estimation with Trust Regions and Double Acquisition Functions, by Giang Ngo and 3 other authors View PDF HTML (experimental) Abstract:Level set estimation (LSE) classifies whether an unknown function's value exceeds a specified threshold for given inputs, a fundamental problem in many real-world applications. In active learning settings with limited initial data, we aim to iteratively acquire informative points to construct an accurate classifier for this task. In high-dimensional spaces, this becomes challenging where the search volume grows exponentially with increasing dimensionality. We propose TRLSE, an algorithm for high-dimensional LSE, which identifies and refines regions near the threshold boundary with dual acquisition functions operating at both global and local levels. We provide a theoretical analysis of TRLSE's accuracy and show its superior sample efficiency against existing methods through extensive evaluations on multiple synthetic and real-world LSE problems. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2602.12391 [cs.LG] (or arXiv:2602.12391v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2602.12391 Focus to learn more arXiv-issued DOI via ...