CyberSec Research

This work introduces a parametric simulation-free reduced order model for incompressible flows undergoing a Hopf bifurcation, leveraging the parametrisation method for invariant manifolds. Unlike data-driven approaches, this method operates directly on the governing equations, eliminating the need for full-order simulations. The proposed model is computed at a single value of the bifurcation parameter yet remains valid over a range of values. The approach systematically constructs an invariant manifold and embedded dynamics, providing an accurate and efficient reduction of the original system. The ability to capture pre-critical steady states, the bifurcation point, and post-critical limit cycle oscillations is demonstrated by a strong agreement between the reduced order model and full order simulations, while achieving significant computational speed-up.

While investment funds publicly disclose their objectives in broad terms, their managers optimize for complex combinations of competing goals that go beyond simple risk-return trade-offs. Traditional approaches attempt to model this through multi-objective utility functions, but face fundamental challenges in specification and parameterization. We propose a generative framework that learns latent representations of fund manager strategies without requiring explicit utility specification. Our approach directly models the conditional probability of a fund's portfolio weights, given stock characteristics, historical returns, previous weights, and a latent variable representing the fund's strategy. Unlike methods based on reinforcement learning or imitation learning, which require specified rewards or labeled expert objectives, our GAN-based architecture learns directly from the joint distribution of observed holdings and market data. We validate our framework on a dataset of 1436 U.S. equity mutual funds. The learned representations successfully capture known investment styles, such as "growth" and "value," while also revealing implicit manager objectives. For instance, we find that while many funds exhibit characteristics of Markowitz-like optimization, they do so with heterogeneous realizations for turnover, concentration, and latent factors. To analyze and interpret the end-to-end model, we develop a series of tests that explain the model, and we show that the benchmark's expert labeling are contained in our model's encoding in a linear interpretable way. Our framework provides a data-driven approach for characterizing investment strategies for applications in market simulation, strategy attribution, and regulatory oversight.

A domain decomposition method for the solution of general variable-coefficient elliptic partial differential equations on regular domains is introduced. The method is based on tessellating the domain into overlapping thin slabs or shells, and then explicitly forming a reduced linear system that connects the different domains. Rank-structure ('H-matrix structure') is exploited to handle the large dense blocks that arise in the reduced linear system. Importantly, the formulation used is well-conditioned, as it converges to a second kind Fredholm equation as the precision in the local solves is refined. Moreover, the dense blocks that arise are far more data-sparse than in existing formulations, leading to faster and more efficient H-matrix arithmetic. To form the reduced linear system, black-box randomized compression is used, taking full advantage of the fact that sparse direct solvers are highly efficient on the thin sub-domains. Numerical experiments demonstrate that our solver can handle oscillatory 2D and 3D problems with as many as 28 million degrees of freedom.

We formulate mold filling in metal casting as a 2D neural operator learning problem that maps geometry and boundary data on an unstructured mesh to time resolved flow quantities, replacing expensive transient CFD. In the proposed method, a graph based encoder aggregates local neighborhood information on the input mesh and encodes geometry and boundary data, a Fourier spectral core operates on a regular latent grid to capture global interactions across the domain, and a graph based decoder projects the latent fields to a target mesh. The model is trained to jointly predict velocity components, pressure, and liquid volume fraction over a fixed rollout horizon and generalizes across different ingate locations and process settings. On held out geometries and inlet conditions, it reproduces large scale advection and the fluid-air interface evolution with localized errors near steep gradients. The mean relative L2 error is about 5% across all fields, and inference is two to three orders of magnitude faster than conventional CFD, enabling design in the loop exploration. Ablation studies show monotonic accuracy degradation under stronger spatial subsampling of input vertices and a smoother decline under temporal subsampling. Halving the training set yields only a small increase in error. These results establish neural operators as accurate and data efficient surrogates for 2D mold filling and enable rapid optimization of gating systems in casting workflows.

Graph Neural Networks (GNNs) have recently been explored as surrogate models for numerical simulations. While their applications in computational fluid dynamics have been investigated, little attention has been given to structural problems, especially for dynamic cases. To address this gap, we introduce the Graph Network-based Structural Simulator (GNSS), a GNN framework for surrogate modeling of dynamic structural problems. GNSS follows the encode-process-decode paradigm typical of GNN-based machine learning models, and its design makes it particularly suited for dynamic simulations thanks to three key features: (i) expressing node kinematics in node-fixed local frames, which avoids catastrophic cancellation in finite-difference velocities; (ii) employing a sign-aware regression loss, which reduces phase errors in long rollouts; and (iii) using a wavelength-informed connectivity radius, which optimizes graph construction. We evaluate GNSS on a case study involving a beam excited by a 50kHz Hanning-modulated pulse. The results show that GNSS accurately reproduces the physics of the problem over hundreds of timesteps and generalizes to unseen loading conditions, where existing GNNs fail to converge or deliver meaningful predictions. Compared with explicit finite element baselines, GNSS achieves substantial inference speedups while preserving spatial and temporal fidelity. These findings demonstrate that locality-preserving GNNs with physics-consistent update rules are a competitive alternative for dynamic, wave-dominated structural simulations.

Although AI drafting tools have gained prominence in patent writing, the systematic evaluation of AI-generated patent content quality represents a significant research gap. To address this gap, We propose to evaluate patents using regulatory compliance, technical coherence, and figure-reference consistency detection modules, and then generate improvement suggestions via an integration module. The framework is validated on a comprehensive dataset comprising 80 human-authored and 80 AI-generated patents from two patent drafting tools. Evaluation is performed on 10,841 total sentences, 8,924 non-template sentences, and 554 patent figures for the three detection modules respectively, achieving balanced accuracies of 99.74%, 82.12%, and 91.2% against expert annotations. Additional analysis was conducted to examine defect distributions across patent sections, technical domains, and authoring sources. Section-based analysis indicates that figure-text consistency and technical detail precision require particular attention. Mechanical Engineering and Construction show more claim-specification inconsistencies due to complex technical documentation requirements. AI-generated patents show a significant gap compared to human-authored ones. While human-authored patents primarily contain surface-level errors like typos, AI-generated patents exhibit more structural defects in figure-text alignment and cross-references.

The proposed system aims to use various machine learning algorithms to enhance financial prediction and generate highly accurate analyses. It introduces an AI-driven platform which offers inflation-analysis, stock market prediction, and E-learning module powered by a chatbot. It has achieved high accuracy where the Inflation Analysis depicts 0.8% MAE, 1.2% RMSE and the Stock Prediction shows 98% and 96% accuracy for Apple and Google stock prices respectively. Key features include historical price trends, inflation rates, short-term future stock prediction, where the data has been extracted using real-world financial datasets. Additionally, the E-learning feature contributes to bridging financial gaps and promoting informed decisions. We have implemented algorithms like linear regression, ARIMA, LSTM where the accuracy has been evaluated using metrics such as MAE, RMSE and the like.

Bayesian methods are particularly effective for addressing inverse problems due to their ability to manage uncertainties inherent in the inference process. However, employing these methods with costly forward models poses significant challenges, especially in the context of non-differentiable models, where the absence of likelihood model gradient information can result in high computational costs. To tackle this issue, we develop a novel Bayesian inference approach based on black box variational inference, utilizing importance sampling to reuse existing simulation model calls in the variational objective gradient estimation, without relying on forward model gradients. The novelty lies in a new batch-sequential sampling procedure, which only requires new model evaluations if the currently available model evaluations fail to yield a suitable approximation of the objective gradient. The resulting approach reduces computational costs by leading to variational parameter updates without requiring new model evaluations when possible, while adaptively increasing the number of model calls per iteration as needed. In combination with its black box nature, this new approach is suitable for inverse problems involving demanding physics-based models that lack model gradients. We demonstrate the efficiency gains of the proposed method compared to its baseline version, sequential Monte Carlo, and Markov-Chain Monte Carlo in diverse benchmarks, ranging from density matching to the Bayesian calibration of a nonlinear electro-chemo-mechanical model for solid-state batteries.

Health indicators (HIs) are central to diagnosing and prognosing the condition of aerospace composite structures, enabling efficient maintenance and operational safety. However, extracting reliable HIs remains challenging due to variability in material properties, stochastic damage evolution, and diverse damage modes. Manufacturing defects (e.g., disbonds) and in-service incidents (e.g., bird strikes) further complicate this process. This study presents a comprehensive data-driven framework that learns HIs via two learning approaches integrated with multi-domain signal processing. Because ground-truth HIs are unavailable, a semi-supervised and an unsupervised approach are proposed: (i) a diversity deep semi-supervised anomaly detection (Diversity-DeepSAD) approach augmented with continuous auxiliary labels used as hypothetical damage proxies, which overcomes the limitation of prior binary labels that only distinguish healthy and failed states while neglecting intermediate degradation, and (ii) a degradation-trend-constrained variational autoencoder (DTC-VAE), in which the monotonicity criterion is embedded via an explicit trend constraint. Guided waves with multiple excitation frequencies are used to monitor single-stiffener composite structures under fatigue loading. Time, frequency, and time-frequency representations are explored, and per-frequency HIs are fused via unsupervised ensemble learning to mitigate frequency dependence and reduce variance. Using fast Fourier transform features, the augmented Diversity-DeepSAD model achieved 81.6% performance, while DTC-VAE delivered the most consistent HIs with 92.3% performance, outperforming existing baselines.

Retrieval-Augmented Generation (RAG) struggles on long, structured financial filings where relevant evidence is sparse and cross-referenced. This paper presents a systematic investigation of advanced metadata-driven Retrieval-Augmented Generation (RAG) techniques, proposing and evaluating a novel, multi-stage RAG architecture that leverages LLM-generated metadata. We introduce a sophisticated indexing pipeline to create contextually rich document chunks and benchmark a spectrum of enhancements, including pre-retrieval filtering, post-retrieval reranking, and enriched embeddings, benchmarked on the FinanceBench dataset. Our results reveal that while a powerful reranker is essential for precision, the most significant performance gains come from embedding chunk metadata directly with text ("contextual chunks"). Our proposed optimal architecture combines LLM-driven pre-retrieval optimizations with these contextual embeddings to achieve superior performance. Additionally, we present a custom metadata reranker that offers a compelling, cost-effective alternative to commercial solutions, highlighting a practical trade-off between peak performance and operational efficiency. This study provides a blueprint for building robust, metadata-aware RAG systems for financial document analysis.

We present a novel neural network architecture for the efficient prediction of sound fields in two and three dimensions. The network is designed to automatically satisfy the Helmholtz equation, ensuring that the outputs are physically valid. Therefore, the method can effectively learn solutions to boundary-value problems in various wave phenomena, such as acoustics, optics, and electromagnetism. Numerical experiments show that the proposed strategy can potentially outperform state-of-the-art methods in room acoustics simulation, in particular in the range of mid to high frequencies.

In this work, we develop a neural network-based, data-driven, decoupled multiscale scheme for the modeling of structured magnetically soft magnetorheological elastomers (MREs). On the microscale, sampled magneto-mechanical loading paths are imposed on a representative volume element containing spherical particles and an elastomer matrix, and the resulting boundary value problem is solved using a mixed finite element formulation. The computed microscale responses are homogenized to construct a database for the training and testing of a macroscopic physics-augmented neural network model. The proposed model automatically detects the material's preferred direction during training and enforces key physical principles, including objectivity, material symmetry, thermodynamic consistency, and the normalization of free energy, stress, and magnetization. Within the range of the training data, the model enables accurate predictions of magnetization, mechanical stress, and total stress. For larger magnetic fields, the model yields plausible results. Finally, we apply the model to investigate the magnetostrictive behavior of a macroscopic spherical MRE sample, which exhibits contraction along the magnetic field direction when aligned with the material's preferred direction.

Aerodynamic simulation of the surface pressure field around objects is crucial for many engineering problems. In recent years, deep neural networks have emerged as an efficient alternative to traditional, computationally expensive CFD simulations for modeling surface pressure fields. However, data scarcity remains a fundamental challenge, limiting the application of neural networks. To address this limitation, we propose to integrate aerodynamic data from multiple subfields and conduct joint training to learn more general field representations. We consolidate five different datasets covering various fields, including automobiles, trains, aircraft, and general shapes. Facing significant data differences across different domains, we propose UniField, which employs a domain-agnostic Transformer module to extract general point cloud features and customizes domain-specific flow-conditioned adapters to adapt to the flow information in different subfields. Despite the fact that aerodynamic data from different subfields are typically governed by different equations, we compare models trained jointly on all data with those trained separately on individual datasets and find that the jointly-trained model commonly demonstrates better performance. This indicates that these data complement each other to help the model learn better flow field representations. These results highlight the potential of UniField as a universal flow field representation model and lay the foundation for broader applications of neural networks in aerodynamic analysis.

High-speed chemically active flows present significant computational challenges due to their disparate space and time scales, where stiff chemistry often dominates simulation time. While modern supercomputing scientific codes achieve exascale performance by leveraging graphics processing units (GPUs), existing GPU-based compressible combustion solvers face critical limitations in memory management, load balancing, and handling the highly localized nature of chemical reactions. To this end, we present a high-performance compressible reacting flow solver built on the AMReX framework and optimized for multi-GPU settings. Our approach addresses three GPU performance bottlenecks: memory access patterns through column-major storage optimization, computational workload variability via a bulk-sparse integration strategy for chemical kinetics, and multi-GPU load distribution for adaptive mesh refinement applications. The solver adapts existing matrix-based chemical kinetics formulations to multigrid contexts. Using representative combustion applications including hydrogen-air detonations and jet in supersonic crossflow configurations, we demonstrate $2-5\times$ performance improvements over initial GPU implementations with near-ideal weak scaling across $1-96$ NVIDIA H100 GPUs. Roofline analysis reveals substantial improvements in arithmetic intensity for both convection ($\sim 10 \times$) and chemistry ($\sim 4 \times$) routines, confirming efficient utilization of GPU memory bandwidth and computational resources.

Post-translational modifications (PTMs) form a combinatorial "code" that regulates protein function, yet deciphering this code - linking modified sites to their catalytic enzymes - remains a central unsolved problem in understanding cellular signaling and disease. We introduce COMPASS-PTM, a mechanism-aware, coarse-to-fine learning framework that unifies residue-level PTM profiling with enzyme-substrate assignment. COMPASS-PTM integrates evolutionary representations from protein language models with physicochemical priors and a crosstalk-aware prompting mechanism that explicitly models inter-PTM dependencies. This design allows the model to learn biologically coherent patterns of cooperative and antagonistic modifications while addressing the dual long-tail distribution of PTM data. Across multiple proteome-scale benchmarks, COMPASS-PTM establishes new state-of-the-art performance, including a 122% relative F1 improvement in multi-label site prediction and a 54% gain in zero-shot enzyme assignment. Beyond accuracy, the model demonstrates interpretable generalization, recovering canonical kinase motifs and predicting disease-associated PTM rewiring caused by missense variants. By bridging statistical learning with biochemical mechanism, COMPASS-PTM unifies site-level and enzyme-level prediction into a single framework that learns the grammar underlying protein regulation and signaling.

For low-frequency electromagnetic problems, where wave-propagation effects can be neglected, eddy current formulations are commonly used as a simplification of the full Maxwell's equations. In this setup, time-domain simulations, needed to capture transient startup responses or nonlinear behavior, are often computationally expensive. We propose a novel tearing and interconnecting approach for eddy currents in time-domain and investigate its scalability.

Artificial intelligence (AI) has emerged as a powerful accelerator of materials discovery, yet most existing models remain problem-specific, requiring additional data collection and retraining for each new property. Here we introduce and validate GATE (Geometrically Aligned Transfer Encoder) -- a generalizable AI framework that jointly learns 34 physicochemical properties spanning thermal, electrical, mechanical, and optical domains. By aligning these properties within a shared geometric space, GATE captures cross-property correlations that reduce disjoint-property bias -- a key factor causing false negatives in multi-criteria screening. To demonstrate its generalizability, GATE -- without any problem-specific reconfiguration -- was directly applied to the discovery of immersion cooling fluids for data centers, a stringent real-world challenge defined by the Open Compute Project (OCP). Screening billions of candidates, GATE identified 92,861 molecules as promising for practical deployment. Four were experimentally or literarily validated, showing strong agreement with wet-lab measurements and performance comparable to or exceeding a commercial coolant. These results establish GATE as a ready-to-use, generalizable AI platform readily applicable across diverse materials discovery tasks.

The low-altitude networks (LANs) integrating unmanned aerial vehicles (UAVs) and high-altitude platforms (HAPs) have become a promising solution for the rising computation demands. However, the uncertain task sizes and high mobility of UAVs pose great challenges to guarantee the quality of service. To address these issues, we propose an LAN architecture where UAVs and HAPs collaboratively provide computation offloading for ground users. Moreover, the uncertainty sets are constructed to characterize the uncertain task size, and a distributionally robust optimization problem is formulated to minimize the worst-case delay by jointly optimizing the offloading decisions and UAV trajectories. To solve the mixed-integer min-max optimization problem, we design the distributionally robust computation offloading and trajectories optimization algorithm. Specifically, the original problem is figured out by iteratively solving the outerlayer and inner-layer problems. The convex outer-layer problem with probability distributions is solved by the optimization toolkit. As for the inner-layer mixed-integer problem, we employ the Benders decomposition. The decoupled master problem concerning the binary offloading decisions is solved by the integer solver, and UAV trajectories in the sub-problem are optimized via the successive convex approximation. Simulation results show the proposed algorithm outperforms traditional optimization methods in balancing the worst-case delay and robustness.