regime detection

This work addresses nonstationary time series characterized by instantaneous effects, nonlinear dynamics, and frequent regime switches by proposing the FlowMSM framework. Under exponential-family noise and a Markov switching mechanism, FlowMSM is the first to achieve simultaneous identifiability of latent regimes and their associated causal structures. The method establishes identifiability theory for a class of Markov-switching models incorporating time-dependent mechanisms, lagged dependencies, and instantaneous effects, and integrates any stationary causal discovery algorithm to recover regime-specific causal graphs. Experiments on both synthetic and real-world financial and economic data demonstrate its effectiveness in accurately identifying latent states and their corresponding causal structures. Notably, non-temporal causal mixture models are encompassed as a special case within this framework.

This study addresses “sudden value misalignment”—an abrupt, significant deviation of large language model (LLM) behavior from human values during fine-tuning, triggered by narrow-domain harmful data. We propose the first interpretable detection framework grounded in natural-language order parameters, integrating statistical distribution shift detection, LLM-based adjudication, and multi-dimensional alignment metrics (e.g., ethics, politics, knowledge). This enables automated identification and decomposed quantification of phase transitions. Key findings reveal that behavioral phase transitions lag behind gradient peaks, underscoring their non-local, emergent nature. Our framework supports fine-grained, cross-domain attribution analysis—precisely quantifying each behavioral dimension’s contribution to overall distributional shift. It thus provides both a theoretical foundation and practical methodology for safe, controllable LLM fine-tuning.

This study addresses the persistent contradictions in empirical assessments of climate and innovation policies by recognizing that the relationship between carbon emissions and economic growth exhibits dynamic heterogeneity during socio-technical transitions—a dimension often overlooked in existing literature. To resolve this, the paper proposes a novel analytical paradigm that operationalizes the theoretical concept of institutional regimes from transition theory by first identifying empirically grounded climate transition mechanisms. Integrating time-varying response analysis, latent variable modeling, and panel data methods within a hybrid econometric–machine learning framework, the authors develop a conditional diagnostic approach. Applying this framework to data from approximately 150 countries over 1991–2022, they successfully uncover distinct mechanisms governing the carbon–economy nexus, each characterized by unique stability and reconfiguration properties, thereby laying a foundation for more precise policy evaluation and forecasting.

To address the cyclic dependency problem arising from the coupling of latent states and nonlinear dynamics in time-series modeling, this paper proposes LaNoLem: a method that models the system as a time-varying dynamical process in a latent space and decouples latent-state inference from dynamics learning via an alternating minimization algorithm. It introduces a fully automated, human-in-the-loop-free complexity regularization criterion to enable adaptive control of model capacity. By jointly optimizing latent-state representation, nonlinear differential equation learning, and dynamics estimation, LaNoLem achieves state-of-the-art accuracy in dynamical system identification. Moreover, it significantly outperforms existing methods on multi-step long-horizon forecasting tasks—particularly for systems exhibiting intricate hidden mechanisms and long-range temporal dependencies.

This work addresses the lack of a general explanation for performance dynamics under varying labeling budgets in active learning, where conventional stage划分 based on fixed label counts fails to generalize. The authors propose a mechanism-driven phase transition theory that reinterprets the budgeting process as shifts in the dominant generalization mechanisms, identifying three distinct phases: data-driven, transitional, and model-driven. By integrating PAC risk decomposition, dynamic reconstruction, measurable proxy metrics, and piecewise regression analysis, they establish a quantifiable framework for phase identification. The study reveals that the efficiency of an acquisition strategy hinges on the alignment between its inductive bias and the prevailing generalization bottleneck. Experiments on natural and medical image datasets validate the three-phase model and demonstrate that self-supervised representations can advance phase transitions, offering a unified foundation for designing phase-aware active learning algorithms.

This work addresses the performance degradation in long-term forecasting of complex systems caused by dataset-level distribution shifts arising from multiple operating regimes and dynamically evolving states. To tackle this challenge, the authors propose NEST, a novel framework that models structural changes through a two-stage dense mixture-of-experts architecture. NEST first performs unsupervised clustering in the moment–entropy space to identify distinct operating regimes, then employs a regime-aware routing mechanism coupled with geometric modulation to dynamically generate expert weights. Each expert functions as a specialized dynamic kernel that captures variable attention patterns specific to its corresponding regime. By explicitly modeling composite, dataset-level operating mechanisms—a capability absent in prior approaches—NEST achieves significant performance gains over existing methods on benchmarks spanning heterogeneous network traffic and physical phenomena.

This work addresses a fundamental limitation in existing adaptive methods, which treat environmental non-stationarity—particularly drift—as mere noise or distributional shift, thereby overlooking the progressive loss of organizational coherence between system and environment over time. To overcome this, the paper introduces the principle of Egregious Drift Regulation (EDR), reframing drift as a regulatory signal of coherence mismatch. EDR enables long-term coherent adaptation by dynamically adjusting the system’s internal structure to maintain, reorganize, or transition its operational mechanisms. Departing from conventional error-minimization objectives, this approach shifts the adaptive goal toward coherence regulation, integrating adaptive control with embodied cognition theory. It realizes a mechanism-centered “emergent machine” architecture that unifies state mechanisms, attractor dynamics, coherence metrics, reconfiguration dynamics, and cross-mechanism memory. The resulting framework offers a principled solution for intelligent systems operating in persistently non-stationary environments, substantially enhancing their long-term functional coherence.

Existing dynamical system reconstruction models exhibit limited out-of-distribution generalization, particularly when extrapolating across critical points. This work identifies three fundamental structural deficiencies underlying this limitation and introduces an improved framework based on topological feature disentanglement and hierarchical modeling. For the first time, the study derives a closed-form theoretical bound characterizing the reliable extrapolation range of such models. The proposed approach enables high-accuracy, zero-shot predictions in unseen dynamical regimes—such as regions straddling bifurcation points—without requiring additional training, thereby substantially enhancing out-of-distribution generalization performance.

This work addresses the heterogeneous training behaviors and failure modes of scientific machine learning (SciML) models across hyperparameter settings, which stem from a lack of unified mechanistic understanding. To bridge this gap, the authors propose a mechanism-aware diagnostic framework that integrates performance metrics, training dynamics, and geometric characteristics of the loss landscape to systematically uncover three prevalent optimization mechanisms in SciML. The framework reveals that optimizer efficacy is highly mechanism-dependent and enables the identification of fine-grained failure modes often invisible to conventional loss landscape analyses. Extensive experiments on prominent SciML architectures—including physics-informed neural networks, neural operators, and neural ordinary differential equations—demonstrate the universality of this tripartite mechanistic structure, offering mechanism-guided design principles for robust SciML optimization.

This work addresses the limited generalization of reinforcement learning policies under unmodeled or time-varying dynamics by proposing a trajectory-outcome-driven implicit dynamics representation that eschews reliance on predefined physical parameters. A task-specific smooth latent space is constructed via semi-supervised contrastive learning, and the authors theoretically establish a monotonic relationship between the regret bound in the target domain and the Lipschitz constant of the trajectory encoder. Leveraging this insight, they enforce Lipschitz constraints to optimize the geometry of the latent space, thereby enhancing robustness. Experiments on MuJoCo benchmarks demonstrate that the proposed method substantially outperforms parameter-centric baselines, effectively handling complex dynamics shifts while improving in-domain stability and interpretability of the latent representation.