arXiv subjects

Eleni Chatzi

Publications and source records attributed to Eleni Chatzi.

At least 19 recordsLinked to original sources

arXiv2026

Reduced-Order Physics-Informed Neural Network with Adaptive Basis Refinement for Structural Identification

Physics-informed neural networks (PINNs) provide a flexible framework for solving forward and inverse problems. However, their direct application to structural dynamics remains limited by high system dimensionality and model-form errors arising from incomplete physics. Reduced-order models (ROMs) can alleviate the dimensionality bottleneck, yet existing PINN-ROM couplings typically rely on fixed reduced subspaces, target forward simulations, or assume complete physics, restricting their use for inverse identification under parametric variability or incomplete system knowledge. To address these limitations, this work proposes a Reduced-Order Physics-Informed Neural Network (RO-PINN) framework with adaptive basis refinement for structural identification under known and incomplete physics. Via projection, reduced governing equations are embedded directly into the PINN loss, facilitating learning in a low-dimensional latent space. An adaptive scheme updates the projection basis during training so that the latent space is progressively realigned with evolving structural parameters or learned residual restoring forces. This realignment reduces basis-mismatch errors and limits their influence on the inferred residual force. The method is validated on a four-story steel frame with nonlinear hysteretic braces under sparse and noisy measurements. Results show parameter identification comparable to or more accurate than Bayesian model updating with lower computational cost in the considered cases, recovery of unmodeled nonlinear restoring forces under incomplete physics, and joint identification of residual restoring forces and structural parameters within the same framework. Overall, RO-PINN provides a unified framework for structural identification by integrating reduced-order modeling, adaptive basis refinement, and physics-informed learning within a single formulation.

cs.CE↗

arXiv2026

Physics-Informed Feature Fusion and Structural Metadata Integration for Transferable Post-Earthquake Damage Classification: Experimental Evaluation and Community-Recovery Implications

Earthquake-induced structural damage assessment remains a key challenge for population-based Structural Health Monitoring (PBSHM), where damage representations must generalize across heterogeneous buildings. Existing vibration-based methods often rely on structure-specific damage-sensitive features (DSFs) and uniform drift thresholds, limiting transferability. This study proposes a physics-informed feature fusion and structural metadata integration framework for post-earthquake damage classification and resilience-oriented assessment. A nonlinear simulation dataset is generated for building populations with varied geometrical and material properties under multiple earthquake scenarios. To obtain consistent labels, damage states are defined through nonlinear pushover analysis and capacity-based thresholds rather than fixed drift limits. Using sparse ground and roof acceleration measurements, physics-informed DSFs are compared with Catch22 and MiniRocket representations under group-wise cross-structure validation with multiple machine-learning classifiers. Results show that physics-informed DSFs outperform generic time-series representations under sparse sensing. Structural metadata further improves robustness by providing context for interpreting response-based DSFs. Modal information offers limited gains in simulations but substantially improves out-of-distribution shaking-table validation, helping bridge simulation-experiment discrepancies.

eess.SP↗

arXiv2026

Are We Making Progress in Multimodal Domain Generalization? A Comprehensive Benchmark Study

Despite the growing popularity of Multimodal Domain Generalization (MMDG) for enhancing model robustness, it remains unclear whether reported performance gains reflect genuine algorithmic progress or are artifacts of inconsistent evaluation protocols. Current research is fragmented, with studies varying significantly across datasets, modality configurations, and experimental settings. Furthermore, existing benchmarks focus predominantly on action recognition, often neglecting critical real-world challenges such as input corruptions, missing modalities, and model trustworthiness. This lack of standardization obscures a reliable assessment of the field's advancement. To address this issue, we introduce MMDG-Bench, the first unified and comprehensive benchmark for MMDG, which standardizes evaluation across six datasets spanning three diverse tasks: action recognition, mechanical fault diagnosis, and sentiment analysis. MMDG-Bench encompasses six modality combinations, nine representative methods, and multiple evaluation settings. Beyond standard accuracy, it systematically assesses corruption robustness, missing-modality generalization, misclassification detection, and out-of-distribution detection. With 7, 402 neural networks trained in total across 95 unique cross-domain tasks, MMDG-Bench yields five key findings: (1) under fair comparisons, recent specialized MMDG methods offer only marginal improvements over ERM baseline; (2) no single method consistently outperforms others across datasets or modality combinations; (3) a substantial gap to upper-bound performance persists, indicating that MMDG remains far from solved; (4) trimodal fusion does not consistently outperform the strongest bimodal configurations; and (5) all evaluated methods exhibit significant degradation under corruption and missing-modality scenarios, with some methods further compromising model trustworthiness.

cs.CV↗

arXiv2026

Towards Interpretable Damage Detection based on Aerodynamic Pressure Measurements

The increasing flexibility of modern large wind turbine blades necessitates cost-efficient and reliable structural monitoring solutions. For this purpose, we propose to use aerodynamic pressure measurements obtained via Aerosense, a novel, non-intrusive and economical sensing system. In former work [Franz et al., 2025], we investigated the potential of aerodynamic pressure measurements for structural damage detection on elastic and aerodynamically loaded structures. An experimental campaign was conducted on a NACA 633418 airfoil mounted on a vertically vibrating cantilever beam within an open wind tunnel. Structural damage was introduced progressively through controlled saw cuts near the beam support. Aerodynamic pressure distributions were recorded under varying inflow conditions and structural states. Based on this data set, we developed a convolutional neural network to detect structural damage and classify its severity using only aerodynamic pressure signals. The results demonstrate that pressure measurements can effectively enable real-time detection and quantification of damage in elastic, beam-like structures subjected to mildly turbulent flow and varying operational conditions. Recognizing the limitations of pure black-box classification, in this study, we further incorporate physics-based insights and explainable machine learning methods to interpret how structural damage influences both the dynamic response and the aerodynamic pressure field. This leads to an enhanced damage detection pipeline, aiming to improve transparency, robustness, and physical consistency in data-driven monitoring of elastic, aerodynamically loaded structures.

eess.SP↗

arXiv2026

PiGGO: Physics-Guided Learnable Graph Kalman Filters for Virtual Sensing of Nonlinear Dynamic Structures under Uncertainty

Digital twins provide a powerful paradigm for diagnostic and prognostic tasks in the monitoring and control of engineered systems; however, their deployment for complex structures remains challenged by model-form uncertainty, arising from unknown nonlinear dynamics, and by sparse sensing. These limitations hinder reliable online state estimation using either purely physics-based or purely data-driven approaches. This work introduces the Physics-Guided Graph Neural ODE (PiGGO) framework, a physics-informed, graph-based Bayesian state estimation approach in which a learned graph neural ordinary differential equation (GNODE) serves as the continuous-time state-transition model within an extended Kalman filter. The graph representation explicitly defines the system state-space, while physics-guided inductive biases encode known structural relationships and constrain the learning of nonlinear dynamics. By integrating graph-native learned dynamics with recursive Bayesian filtering, the proposed PiGGO framework enables online virtual sensing and uncertainty-aware state estimation for nonlinear systems with unknown model form, while maintaining generalisation across topologically similar structures. Numerical case studies demonstrate improved robustness to model uncertainty and measurement noise, outperforming both open-loop graph neural models and conventional filtering approaches in online prediction tasks.

cs.LG↗

arXiv2026

Disentangling Damage from Operational Variability: A Label-Free Self-Supervised Representation Learning Framework for Output-Only Structural Damage Identification

Damage identification is a core task in structural health monitoring. In practice, however, its reliability is often compromised by confounding non-damage effects, such as variations in excitation and environmental conditions, which can induce changes comparable to or larger than those caused by structural damage. To address this challenge, this study proposes a self-supervised label-free disentangled representation learning framework for robust vibration-based structural damage identification. The proposed framework employs an autoencoder with two latent representations to learn directly from raw vibration acceleration signals. A self-supervised invariance regularization, implemented via Variance-Invariance-Covariance Regularization (VICReg), is imposed on one latent representation using baseline data where structural damage is assumed constant but operational and environmental conditions vary. In addition, a frequency-domain constraint is introduced to enforce agreement between the power spectral density reconstructed from the latent representation and that computed from the corresponding input time series. Together, these mechanisms promote disentanglement, enabling the learned representation to be sensitive to damage-related characteristics while remaining invariant to nuisance variability. The framework is trained in a fully end-to-end and label-free manner, requiring no prior information on damage, excitation, or environmental conditions, making it well-suited for real-world applications. Its effectiveness is validated on two distinct real-world vibration datasets, including a bridge and a gearbox. The results demonstrate robustness to operational variability, strong generalization capability, and good performance in both damage detection and quantification.

cs.LG↗

arXiv2026

Neuro-Symbolic ODE Discovery with Latent Grammar Flow

Understanding natural and engineered systems often relies on symbolic formulations, such as differential equations, which provide interpretability and transferability beyond black-box models. We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data. Domain knowledge and constraints, such as stability, can be either embedded into the rules or used as conditional predictors.

cs.LG↗

arXiv2026

Multimodal Learning for Arcing Detection in Pantograph-Catenary Systems

The pantograph-catenary interface is essential for ensuring uninterrupted and reliable power delivery in electrified rail systems. However, electrical arcing at this interface poses serious risks, including accelerated wear of contact components, degraded system performance, and potential service disruptions. Detecting arcing events at the pantograph-catenary interface is challenging due to their transient nature, noisy operating environment, data scarcity, and the difficulty of distinguishing arcs from other similar transient phenomena. To address these challenges, we propose a novel multimodal framework that combines high-resolution image data with force measurements to more accurately and robustly detect arcing events. First, we construct two arcing detection datasets comprising synchronized visual and force measurements. One dataset is built from data provided by the Swiss Federal Railways (SBB), and the other is derived from publicly available videos of arcing events in different railway systems and synthetic force data that mimic the characteristics observed in the real dataset. Leveraging these datasets, we propose MultiDeepSAD, an extension of the DeepSAD algorithm for multiple modalities with a new loss formulation. Additionally, we introduce tailored pseudo-anomaly generation techniques specific to each data type, such as synthetic arc-like artifacts in images and simulated force irregularities, to augment training data and improve the discriminative ability of the model. Through extensive experiments and ablation studies, we demonstrate that our framework significantly outperforms baseline approaches, exhibiting enhanced sensitivity to real arcing events even under domain shifts and limited availability of real arcing observations.

cs.CV↗

arXiv2025

A Mechanistic Analysis of Transformers for Dynamical Systems

Transformers are increasingly adopted for modeling and forecasting time-series, yet their internal mechanisms remain poorly understood from a dynamical systems perspective. In contrast to classical autoregressive and state-space models, which benefit from well-established theoretical foundations, Transformer architectures are typically treated as black boxes. This gap becomes particularly relevant as attention-based models are considered for general-purpose or zero-shot forecasting across diverse dynamical regimes. In this work, we do not propose a new forecasting model, but instead investigate the representational capabilities and limitations of single-layer Transformers when applied to dynamical data. Building on a dynamical systems perspective, we interpret causal self-attention as a linear, history-dependent recurrence and analyze how it processes temporal information. Through a series of linear and nonlinear case studies, we identify distinct operational regimes. For linear systems, we show that in the single-head attention-only setting, the convexity constraint imposed by softmax attention restricts the class of autoregressive operators that can be represented, leading to oversmoothing when the target dynamics require mixed-sign lag coefficients. For nonlinear systems under partial observability, attention instead acts as an adaptive delay-embedding mechanism, enabling effective state reconstruction when sufficient temporal context and latent dimensionality are available. These results help bridge empirical observations with classical dynamical systems theory, providing insight into when and why Transformers succeed or fail as models of dynamical systems.

cs.LG↗

arXiv2025

Causal Convolutional Neural Networks as Finite Impulse Response Filters

This study investigates the behavior of Causal Convolutional Neural Networks (CNNs) with quasi-linear activation functions when applied to time-series data characterized by multimodal frequency content. We demonstrate that, once trained, such networks exhibit properties analogous to Finite Impulse Response (FIR) filters, particularly when the convolutional kernels are of extended length exceeding those typically employed in standard CNN architectures. Causal CNNs are shown to capture spectral features both implicitly and explicitly, offering enhanced interpretability for tasks involving dynamic systems. Leveraging the associative property of convolution, we further show that the entire network can be reduced to an equivalent single-layer filter resembling an FIR filter optimized via least-squares criteria. This equivalence yields new insights into the spectral learning behavior of CNNs trained on signals with sparse frequency content. The approach is validated on both simulated beam dynamics and real-world bridge vibration datasets, underlining its relevance for modeling and identifying physical systems governed by dynamic responses.

cs.LG↗

arXiv2025

A GAN-Based Framework for Generating STFT Spectrograms of Rare Acoustic Events in Structural Health Monitoring

Structural Health Monitoring plays a crucial role in ensuring the safety, reliability, and longevity of bridge infrastructures through early damage detection. Although recent advances in deep learning-based models have enabled automated event detection, their performance is often limited by data scarcity, environmental noise, and class imbalance. To address these challenges, this study introduces a customized Generative Adversarial Network model, STFTSynth, designed particularly for generating short-time Fourier transform spectrograms derived from acoustic event signals. In contrast to augmentation techniques such as MixUp, generative adversarial networks can synthesize spectrograms that visually and statistically resemble real event representations, providing a basis for representation-level dataset enrichment. The proposed model integrates dense residual blocks for spatial consistency with bidirectional gated recurrent units for temporal dependency modeling. Model performance is evaluated against three baseline generative models using qualitative inspection and quantitative metrics, including Structural Similarity Index Measure, Peak Signal-to-Noise Ratio, and Frechet Inception Distance. Results show that STFTSynth outperforms baseline models, producing high-resolution, temporally consistent spectrograms that align closely with real-world data. These findings highlight the potential of GAN-based spectrogram synthesis for representation-level enrichment of rare acoustic-event datasets in bridge monitoring, particularly when real examples such as prestressing wire breakage are limited.

cs.CE↗

arXiv2025

On the Rate of Convergence of Kolmogorov-Arnold Network Regression Estimators

Kolmogorov-Arnold Networks (KANs) approximate multivariate functions by composing univariate transformations through additive or multiplicative aggregation. We establish convergence guarantees for KANs whose univariate components are B-splines. The least-squares estimator over the KAN spline sieve attains the rate $O((\log n / n)^{2r/(2r+1)})$, uniformly over a ball of regression functions admitting a KAN representation with univariate components of Sobolev smoothness $r$; a matching lower bound of order $n^{-2r/(2r+1)}$ shows this is minimax optimal up to the logarithmic factor, which we trace to the nonlinearity of the sieve rather than to the architecture. The rate is free of the ambient dimension $d$; this dimension-free exponent reflects the assumed KAN structure of the target, not an escape from the minimax rate $n^{-2r/(2r+d)}$ on Sobolev classes over $[0,1]^d$. We derive a knot-selection rule, show that penalized selection over a dyadic knot grid attains the rate adaptively in the unknown smoothness, and show that univariate components are not identifiable under centering alone, so consistency of the fit does not imply consistency of the components. On targets of exactly known smoothness the fitted risk exponent is at least as steep as the bound in every configuration, and the predicted knot scaling and $k^{-r}$ approximation decay are checked directly.

cs.LG↗

arXiv2025

Geometric Autoencoder Priors for Bayesian Inversion: Learn First Observe Later

Uncertainty Quantification (UQ) is paramount for inference in engineering. A common inference task is to recover full-field information of physical systems from a small number of noisy observations, a usually highly ill-posed problem. Sharing information from multiple distinct yet related physical systems can alleviate this ill-posedness. Critically, engineering systems often have complicated variable geometries prohibiting the use of standard multi-system Bayesian UQ. In this work, we introduce Geometric Autoencoders for Bayesian Inversion (GABI), a framework for learning geometry-aware generative models of physical responses that serve as highly informative geometry-conditioned priors for Bayesian inversion. Following a ''learn first, observe later'' paradigm, GABI distills information from large datasets of systems with varying geometries, without requiring knowledge of governing PDEs, boundary conditions, or observation processes, into a rich latent prior. At inference time, this prior is seamlessly combined with the likelihood of a specific observation process, yielding a geometry-adapted posterior distribution. Our proposed framework is architecture-agnostic. A creative use of Approximate Bayesian Computation (ABC) sampling yields an efficient implementation that utilizes modern GPU hardware. We test our method on: steady-state heat over rectangular domains; Reynolds-Averaged Navier-Stokes (RANS) flow around airfoils; Helmholtz resonance and source localization on 3D car bodies; RANS airflow over terrain. We find: the predictive accuracy to be comparable to deterministic supervised learning approaches in the restricted setting where supervised learning is applicable; UQ to be well calibrated and robust on challenging problems with complex geometries.

stat.ML↗

arXiv2025

Quantifying the Value of Seismic Structural Health Monitoring for post-earthquake recovery of electric power system in terms of resilience enhancement

Post-earthquake recovery of electric power networks (EPNs) is critical to community resilience. Traditional recovery processes often rely on prolonged and imprecise manual inspections for damage diagnosis, leading to suboptimal repair prioritization and extended service disruptions. Seismic Structural Health Monitoring (SSHM) offers the potential to expedite recovery by enabling more accurate and timely damage assessment. However, SSHM deployment incurs costs, and its system-level resilience benefit remains underexplored. This study proposes a probabilistic simulation framework to quantify the value of SSHM for enhancing EPN resilience. The framework includes seismic damage modeling based on network configuration, hazard intensity, fragility functions, and damage-functionality mappings, combined with recovery simulations incorporating resource constraints, repair and transfer durations. System functionality is evaluated using graph-based island detection and optimal power flow analysis. Resilience is quantified via the Lack of Resilience (LoR) metric derived from the functionality restoration curve. SSHM is incorporated by altering the quality of damage information used in repair scheduling. Different monitoring scenarios (e.g., no-SSHM baseline, partial SSHM, full SSHM with various accuracies) are modeled using confusion matrices to simulate damage misclassification. Results show that improved damage awareness via SSHM significantly accelerates recovery and reduces LoR by up to 21%. This work supports evidence-based decisions for SSHM deployment in critical infrastructure.

eess.SY↗

arXiv2025

A Multi-Model Probabilistic Framework for Seismic Risk Assessment and Retrofit Planning of Electric Power Networks

Electric power networks are critical lifelines, and their disruption during earthquakes can lead to severe cascading failures and significantly hinder post-disaster recovery. Enhancing their seismic resilience requires identifying and strengthening vulnerable components in a cost-effective and system-aware manner. However, existing studies often overlook the systemic behavior of power networks under seismic loading. Common limitations include isolated component analyses that neglect network-wide interdependencies, oversimplified damage models assuming binary states or damage independence, and the exclusion of electrical operational constraints. These simplifications can result in inaccurate risk estimates and inefficient retrofit decisions. This study proposes a multi-model probabilistic framework for seismic risk assessment and retrofit planning of electric power systems. The approach integrates: (1) regional seismic hazard characterization with ground motion prediction and spatial correlation models; (2) component-level damage analysis using fragility functions and multi-state damage-functionality mappings; (3) system-level cascading impact evaluation through graph-based island detection and constrained optimal power flow analysis; and (4) retrofit planning via heuristic optimization to minimize expected annual functionality loss (EAFL) under budget constraints. Uncertainty is propagated throughout the framework using Monte Carlo simulation. The methodology is demonstrated on the IEEE 24-bus Reliability Test System, showcasing its ability to capture cascading failures, identify critical components, and generate effective retrofit strategies. Results underscore the potential of the framework as a scalable, data-informed decision-support tool for enhancing the seismic resilience of power infrastructure.

eess.SY↗

arXiv2025

Adapting Vision-Language Models Without Labels: A Comprehensive Survey

Vision-Language Models (VLMs) have demonstrated remarkable generalization capabilities across a wide range of tasks. However, their performance often remains suboptimal when directly applied to specific downstream scenarios without task-specific adaptation. To enhance their utility while preserving data efficiency, recent research has increasingly focused on unsupervised adaptation methods that do not rely on labeled data. Despite the growing interest in this area, there remains a lack of a unified, task-oriented survey dedicated to unsupervised VLM adaptation. To bridge this gap, we present a comprehensive and structured overview of the field. We propose a taxonomy based on the availability and nature of unlabeled visual data, categorizing existing approaches into four key paradigms: Data-Free Transfer (no data), Unsupervised Domain Transfer (abundant data), Episodic Test-Time Adaptation (batch data), and Online Test-Time Adaptation (streaming data). Within this framework, we analyze core methodologies and adaptation strategies associated with each paradigm, aiming to establish a systematic understanding of the field. Additionally, we review representative benchmarks across diverse applications and highlight open challenges and promising directions for future research. An actively maintained repository of relevant literature is available at https://github.com/tim-learn/Awesome-LabelFree-VLMs.

cs.LG↗

arXiv2025

Beyond Static Models: Hypernetworks for Adaptive and Generalizable Forecasting in Complex Parametric Dynamical Systems

Dynamical systems play a key role in modeling, forecasting, and decision-making across a wide range of scientific domains. However, variations in system parameters, also referred to as parametric variability, can lead to drastically different model behavior and output, posing challenges for constructing models that generalize across parameter regimes. In this work, we introduce the Parametric Hypernetwork for Learning Interpolated Networks (PHLieNet), a framework that simultaneously learns: (a) a global mapping from the parameter space to a nonlinear embedding and (b) a mapping from the inferred embedding to the weights of a dynamics propagation network. The learned embedding serves as a latent representation that modulates a base network, termed the hypernetwork, enabling it to generate the weights of a target network responsible for forecasting the system's state evolution conditioned on the previous time history. By interpolating in the space of models rather than observations, PHLieNet facilitates smooth transitions across parameterized system behaviors, enabling a unified model that captures the dynamic behavior across a broad range of system parameterizations. The performance of the proposed technique is validated in a series of dynamical systems with respect to its ability to extrapolate in time and interpolate and extrapolate in the parameter space, i.e., generalize to dynamics that were unseen during training. Our approach outperforms state-of-the-art baselines in both short-term forecast accuracy and in capturing long-term dynamical features such as attractor statistics.

cs.LG↗

arXiv2025

Structured Kolmogorov-Arnold Neural ODEs for Interpretable Learning and Symbolic Discovery of Nonlinear Dynamics

Understanding and modeling nonlinear dynamical systems is a fundamental challenge across science and engineering. Deep learning has shown remarkable potential for capturing complex system behavior, yet achieving models that are both accurate and physically interpretable remains difficult. To address this, we propose Structured Kolmogorov-Arnold Neural ODEs (SKANODEs), a framework that integrates structured state-space modeling with Kolmogorov-Arnold Networks (KANs). Within a Neural ODE architecture, SKANODE employs a fully trainable KAN as a universal function approximator to perform virtual sensing, recovering latent states that correspond to interpretable physical quantities such as displacements and velocities. Leveraging KAN's symbolic regression capability, SKANODE then extracts compact, interpretable expressions for the system's governing dynamics. Experiments on two canonical nonlinear oscillators and a real-world F-16 ground vibration dataset demonstrate that SKANODE reliably recovers physically meaningful latent displacement and velocity trajectories from acceleration measurements, identifies the correct governing nonlinearities--including the cubic stiffness in the Duffing oscillator and the nonlinear damping structure in the Van der Pol oscillator--and reveals hysteretic signatures in the F-16 interface dynamics through structured latent phase portraits and an interpretable symbolic model. Across all three cases, SKANODE provides more accurate and robust predictions than black-box NODE baselines and classical ARX and NARX identification, while producing equation-level descriptions of the learned nonlinear dynamics.

cs.LG↗