mlsanityData to 14 Oct 2026

Noise-robust Contrastive Learning for Critical Transition Detection in Dynamical Systems

arXiv 2512.12523 ↗First posted 14 Dec 2025cs.LG cs.AI physics.comp-phcited by 0

Not on the map. Why?

Abstract (arXiv)

Detecting critical transitions in complex, noisy time-series data is a fundamental challenge across science and engineering. Such transitions may be anticipated by the emergence of a low-dimensional order parameter, whose signature is often masked by high-amplitude stochastic variability. Standard contrastive learning approaches based on deep neural networks, while promising for detecting critical transitions, are often overparameterized and sensitive to irrelevant noise, leading to inaccurate identification of critical points. To address these limitations, we propose a neural network architecture, constructed using singular value decomposition technique, together with a strictly semi-orthogonality-constrained training algorithm, to enhance the performance of traditional contrastive learning. Extensive experiments demonstrate that the proposed method matches the performance of traditional contrastive learning techniques in identifying critical transitions, yet is considerably more lightweight and markedly more resistant to noise.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 24 of its 57 references that are in the map. First posted on arXiv 14 Dec 2025.

Builds on

24 of 57 references.

References · 24

  1. 2019Unsupervised Scalable Representation Learning for Multivariate Time Seriescited by 269
  2. 2022Selective-Supervised Contrastive Learning with Noisy Labelscited by 112
  3. 2020Multi-Objective Interpolation Training for Robustness to Label Noisecited by 86
  4. 2023Manifold learning: what, how, and whycited by 68
  5. 2020Machine learning prediction of critical transition and system collapsecited by 54
  6. 2022Robust Contrastive Learning against Noisy Viewscited by 45
  7. 2022Hyperbolic Graph Neural Networks: A Review of Methods and Applicationscited by 40
  8. 2018The loss landscape of overparameterized neural networkscited by 39
  9. 2023Symmetry Induces Structure and Constraint of Learningcited by 25
  10. 2014From Manifold to Manifold: Geometry-Aware Dimensionality Reduction for SPD Matricescited by 22
  11. 2022Deep Learning of Causal Structures in High Dimensionscited by 20
  12. 2025An overview of condensation phenomenon in deep learningcited by 17
12 more
  1. 2025Parameter Symmetry Potentially Unifies Deep Learning Theorycited by 12
  2. 2022Deep Causal Learning: Representation, Discovery and Inferencecited by 11
  3. 2022Universal Early Warning Signals of Phase Transitions in Climate Systemscited by 10
  4. 2024SympGNNs: Symplectic Graph Neural Networks for identifiying high-dimensional Hamiltonian systems and node classificationcited by 4
  5. 2022A simple framework for contrastive learning phases of mattercited by 3
  6. 2023A predictive physics-aware hybrid reduced order model for reacting flowscited by 3
  7. 2024Regularized Gauss-Newton for Optimizing Overparameterized Neural Networkscited by 3
  8. 2025Low-Rank Matrix Approximation for Neural Network Compressioncited by 3
  9. 2025An Inclusive Theoretical Framework of Robust Supervised Contrastive Loss against Label Noisecited by 2
  10. 2025Ultralow-dimensionality reduction for identifying critical transitions by spatial-temporal PCAcited by 2
  11. 2025Low-Rank Tensor Decompositions for the Theory of Neural Networkscited by 2
  12. 2025How Does Overparameterization Affect Machine Unlearning of Deep Neural Networks?cited by 1

Built on by

Not cited yet.