Conformal Online Learning of Deep Koopman Linear Embeddings
Ben Gao, Jordan Patracone, Stéphane Chrétien, Olivier Alata
Abstract
We introduce Conformal Online Learning of Koopman embeddings (COLoKe), a novel framework for adaptively updating Koopman-invariant representations of nonlinear dynamical systems from streaming data. Our modeling approach combines deep feature learning with multistep prediction consistency in the lifted space, where the dynamics evolve linearly. To prevent overfitting, COLoKe employs a conformal-style mechanism that shifts the focus from evaluating the conformity of new states to assessing the consistency of the current Koopman model. Updates are triggered only when the current model's prediction error exceeds a dynamically calibrated threshold, allowing selective refinement of the Koopman operator and embedding. Empirical results on benchmark dynamical systems demonstrate the effectiveness of COLoKe in maintaining long-term predictive accuracy while significantly reducing unnecessary updates and avoiding overfitting.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Improved Online Conformal Prediction via Strongly Adaptive Online Learning cites Adaptive Conformal Inference Under Distribution Shift
- Conformal PID Control for Time Series Prediction cites Improved Online Conformal Prediction via Strongly Adaptive Online Learning
- Conformal PID Control for Time Series Prediction cites Adaptive Conformal Inference Under Distribution Shift
Builds on9
- Informer: Beyond Efficient Transformer for Long Sequence Time-Series ForecastingHaoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang et al.AAAI 2021 · 7,289 citations
- Adaptive Conformal Inference Under Distribution ShiftIsaac Gibbs, Emmanuel J. CandèsNeurIPS 2021 · 665 citations
- Conformal PID Control for Time Series PredictionAnastasios Angelopoulos, Emmanuel J. Candès, Ryan J. TibshiraniNeurIPS 2023 · 164 citations
- Learning Dynamical Systems via Koopman Operator Regression in Reproducing Kernel Hilbert SpacesVladimir Kostic, Pietro Novelli, Andreas Maurer, Carlo Ciliberto et al.NeurIPS 2022 · 109 citations
- Improved Online Conformal Prediction via Strongly Adaptive Online LearningAadyot Bhatnagar, Huan Wang, Caiming Xiong, Yu BaiICML 2023 · 87 citations
Related papers
- Online Learning and Control of Complex Dynamical Systems from Sensory InputOumayma Bounou, Jean Ponce, Justin CarpentierNeurIPS 2021 · 10 citations
- AdaKoop: Efficient Modeling of Nonlinear Dynamics from Nonstationary Data Streams with Koopman Operator RegressionNaoki Chihara, Ren Fujiwara, Yasuko Matsubara, Yasushi SakuraiKDD 2026
- Efficient Parametric SVD of Koopman Operator for Stochastic Dynamical SystemsMinchan Jeong, Jongha Ryu, Se-Young Yun, Gregory W. WornellNeurIPS 2025 · 6 citations
- Forecasting Sequential Data Using Consistent Koopman AutoencodersOmri Azencot, N. Benjamin Erichson, Vanessa Lin, Michael W. MahoneyICML 2020 · 203 citations
- Koopa: Learning Non-stationary Time Series Dynamics with Koopman PredictorsYong Liu, Chenyu Li, Jianmin Wang, Mingsheng LongNeurIPS 2023 · 284 citations
