Understanding recurrent neural networks using nonequilibrium response theory Journal of Machine Learning Research, 2021
Noisy Recurrent Neural Networks Advances in Neural Information Processing Systems, 2021
Predicting critical transitions in multiscale dynamical systems using reservoir computing Soon Hoe Lim, Ludovico Theo Giorgini, Woosok Moon, J. S. Wettlaufer Chaos, 2020 We study the problem of predicting rare critical transition events for a class of slow-fast nonlinear dynamical systems. The state of the system of interest is described by a slow process, whereas a faster process drives its evolution and induces critical transitions. By taking advantage of recent advances in reservoir computing, we present a data-driven method to predict the future evolution of the state. We show that our method is capable of predicting a critical transition event at least several numerical time steps in advance. We demonstrate the success as well as the limitations of our method using numerical experiments on three examples of systems, ranging from low dimensional to high dimensional. We discuss the mathematical and broader implications of our results.
Homogenization for Generalized Langevin Equations with Applications to Anomalous Diffusion Soon Hoe Lim, Jan Wehr, Maciej Lewenstein Annales Henri Poincare, 2020 We study homogenization for a class of generalized Langevin equations (GLEs) with state-dependent coefficients and exhibiting multiple time scales. In addition to the small mass limit, we focus on homogenization limits, which involve taking to zero the inertial time scale and, possibly, some of the memory time scales and noise correlation time scales. The latter are meaningful limits for a class of GLEs modeling anomalous diffusion. We find that, in general, the limiting stochastic differential equations for the slow degrees of freedom contain non-trivial drift correction terms and are driven by non-Markov noise processes. These results follow from a general homogenization theorem stated and proven here. We illustrate them using stochastic models of particle diffusion.
Sharpen Your Flow: Sharpness-Aware Sampling for Flow Matching A Gupta, SH Lim, A Yu, NB Erichson arXiv preprint arXiv:2605.11547 , 2026 2026
Is Flow Matching Just Trajectory Replay for Sequential Data? SH Lim, S Lin, MW Mahoney, NB Erichson arXiv preprint arXiv:2602.08318 , 2026 2026 Citations: 1
A Kinetic-Energy Perspective of Flow Matching Z Li, H Hu, SH Lim, X Li, F Gao, E Diao, Z Ding, M Vazirgiannis, ... arXiv preprint arXiv:2602.07928 , 2026 2026 Citations: 2
On The Hidden Biases of Flow Matching Samplers SH Lim arXiv preprint arXiv:2512.16768v3 , 2025 2025
EnfoPath: Energy-Informed Analysis of Generative Trajectories in Flow Matching Z Li, B Dai, H Hu, H Boström, SH Lim arXiv preprint arXiv:2511.19087 , 2025 2025 Citations: 1
Flex: A backbone for diffusion-based modeling of spatio-temporal physical systems NB Erichson, V Mikuni, D Lyu, Y Gao, O Azencot, SH Lim, MW Mahoney arXiv preprint arXiv:2505.17351 , 2025 2025 Citations: 5
Tuning frequency bias of state space models A Yu, D Lyu, SH Lim, MW Mahoney, NB Erichson ICLR 2025 (Spotlight) , 2025 2025 Citations: 16
Elucidating the design choice of probability paths in flow matching for forecasting SH Lim, Y Wang, A Yu, E Hart, MW Mahoney, XS Li, NB Erichson Transactions on Machine Learning Research , 2024 2024 Citations: 10
Stochastic Processes: From Classical to Quantum SH Lim arXiv preprint arXiv:2407.04005 , 2024 2024 Citations: 2
NoisyMix: Boosting Model Robustness to Common Corruptions NB Erichson, SH Lim, W Xu, F Utrera, Z Cao, MW Mahoney Proc. of the 27th International Conference on AISTATS , 2024 2024 Citations: 32
Gated recurrent neural networks with weighted time-delay feedback NB Erichson, SH Lim, MW Mahoney Proc. of the 28th International Conference on AISTATS , 2022 2022 Citations: 9
Chaotic Regularization and Heavy-Tailed Limits for Deterministic Gradient Descent SH Lim, Y Wan, U Şimşekli Advances in Neural Information Processing Systems 35 , 2022 2022 Citations: 18
Noisymix: Boosting robustness by combining data augmentations, stability training, and noise injections NB Erichson, SH Lim, F Utrera, W Xu, Z Cao, MW Mahoney arXiv preprint arXiv:2202.01263 1 , 2022 2022 Citations: 28
Noisy Feature Mixup SH Lim, NB Erichson, F Utrera, W Xu, MW Mahoney Proc. of the 2022 ICLR Conference , 2021 2021 Citations: 64
Anomalous thermodynamics in homogenized generalized Langevin systems SH Lim Journal of Physics A: Mathematical and Theoretical 54 (15), 155001 , 2021 2021 Citations: 4
Noisy Recurrent Neural Networks SH Lim, NB Erichson, L Hodgkinson, MW Mahoney Advances in Neural Information Processing Systems 34 , 2021 2021 Citations: 87
Modeling the El Nino Southern Oscillation with Neural Differential Equations LT Giorgini, SH Lim, W Moon, N Chen, JS Wettlaufer ICML 2021 Time Series Workshop , 2021 2021 Citations: 5
Understanding Recurrent Neural Networks Using Nonequilibrium Response Theory SH Lim Journal of Machine Learning Research 22, 47:1-47:48 , 2021 2021 Citations: 24
Predicting critical transitions in multiscale dynamical systems using reservoir computing SH Lim, LT Giorgini, W Moon, JS Wettlaufer Chaos: An Interdisciplinary Journal of Nonlinear Science 30 (12) , 2020 2020 Citations: 45
Homogenization for generalized Langevin equations with applications to anomalous diffusion SH Lim, J Wehr, M Lewenstein Annales Henri Poincaré 21, 1813–1871 , 2020 2020 Citations: 22
MOST CITED SCHOLAR PUBLICATIONS
Bose polaron as an instance of quantum Brownian motion A Lampo, SH Lim, MÁ García-March, M Lewenstein Quantum 1, 30 , 2017 2017 Citations: 100
Noisy Recurrent Neural Networks SH Lim, NB Erichson, L Hodgkinson, MW Mahoney Advances in Neural Information Processing Systems 34 , 2021 2021 Citations: 87
Noisy Feature Mixup SH Lim, NB Erichson, F Utrera, W Xu, MW Mahoney Proc. of the 2022 ICLR Conference , 2021 2021 Citations: 64
Predicting critical transitions in multiscale dynamical systems using reservoir computing SH Lim, LT Giorgini, W Moon, JS Wettlaufer Chaos: An Interdisciplinary Journal of Nonlinear Science 30 (12) , 2020 2020 Citations: 45
Lindblad model of quantum Brownian motion A Lampo, SH Lim, J Wehr, P Massignan, M Lewenstein Physical Review A 94 (4), 042123 , 2016 2016 Citations: 34
NoisyMix: Boosting Model Robustness to Common Corruptions NB Erichson, SH Lim, W Xu, F Utrera, Z Cao, MW Mahoney Proc. of the 27th International Conference on AISTATS , 2024 2024 Citations: 32
Noisymix: Boosting robustness by combining data augmentations, stability training, and noise injections NB Erichson, SH Lim, F Utrera, W Xu, Z Cao, MW Mahoney arXiv preprint arXiv:2202.01263 1 , 2022 2022 Citations: 28
Functionals in stochastic thermodynamics: how to interpret stochastic integrals S Bo, SH Lim, R Eichhorn Journal of Statistical Mechanics: Theory and Experiment 2019 (8), 084005 , 2019 2019 Citations: 27
Understanding Recurrent Neural Networks Using Nonequilibrium Response Theory SH Lim Journal of Machine Learning Research 22, 47:1-47:48 , 2021 2021 Citations: 24
Homogenization for generalized Langevin equations with applications to anomalous diffusion SH Lim, J Wehr, M Lewenstein Annales Henri Poincaré 21, 1813–1871 , 2020 2020 Citations: 22
Homogenization for a class of generalized Langevin equations with an application to thermophoresis SH Lim, J Wehr Journal of Statistical Physics 174 (3), 656-691 , 2019 2019 Citations: 20
Chaotic Regularization and Heavy-Tailed Limits for Deterministic Gradient Descent SH Lim, Y Wan, U Şimşekli Advances in Neural Information Processing Systems 35 , 2022 2022 Citations: 18
Tuning frequency bias of state space models A Yu, D Lyu, SH Lim, MW Mahoney, NB Erichson ICLR 2025 (Spotlight) , 2025 2025 Citations: 16
On the small mass limit of quantum Brownian motion with inhomogeneous damping and diffusion SH Lim, J Wehr, A Lampo, MÁ García-March, M Lewenstein Journal of Statistical Physics 170 (2), 351-377 , 2018 2018 Citations: 16
Precursors to rare events in stochastic resonance LT Giorgini, SH Lim, W Moon, JS Wettlaufer Europhysics letters 129 (4), 40003 , 2020 2020 Citations: 13
Elucidating the design choice of probability paths in flow matching for forecasting SH Lim, Y Wang, A Yu, E Hart, MW Mahoney, XS Li, NB Erichson Transactions on Machine Learning Research , 2024 2024 Citations: 10
Gated recurrent neural networks with weighted time-delay feedback NB Erichson, SH Lim, MW Mahoney Proc. of the 28th International Conference on AISTATS , 2022 2022 Citations: 9
Flex: A backbone for diffusion-based modeling of spatio-temporal physical systems NB Erichson, V Mikuni, D Lyu, Y Gao, O Azencot, SH Lim, MW Mahoney arXiv preprint arXiv:2505.17351 , 2025 2025 Citations: 5
Modeling the El Nino Southern Oscillation with Neural Differential Equations LT Giorgini, SH Lim, W Moon, N Chen, JS Wettlaufer ICML 2021 Time Series Workshop , 2021 2021 Citations: 5
Anomalous thermodynamics in homogenized generalized Langevin systems SH Lim Journal of Physics A: Mathematical and Theoretical 54 (15), 155001 , 2021 2021 Citations: 4
