arXiv subjects

Richard Tran

Publications and source records attributed to Richard Tran.

10 recordsLinked to original sources

arXiv2026

Leveraging LLMs for Unstructured Claims Data Analysis

Actuaries rely primarily on structured numerical data for reserving and ratemaking, while valuable predictive information in unstructured text including medical records, adjuster notes, and call transcripts remains largely unused. Manual processing of these documents is time-consuming, inconsistent across reviewers, and unscalable. We present a proof-of-concept framework using large language models (LLMs) to extract structured actuarial variables from unstructured claims data. We implement a two-stage processing architecture separating document-level extraction (Stage 1) from claim-level synthesis (Stage 2). A modular four-script Python pipeline processes synthetic FHIR-based claims data and real claims documents, extracting 36 actuarial variables across reserving, ratemaking, and claims management categories. We validate 14 core variables using two independent clinical expert reviewers scoring 20 synthetic claims on a five-point Likert rubric, achieving mean scores above 4.0 and a weighted kappa of 0.53. Integration with chain ladder reserving demonstrates practical actuarial value: severity-segmented analysis reduced reserve estimation error from 6.5% to 4.0%. The open-source implementation includes audit trails and confidence scoring, providing a replicable foundation for LLM-based actuarial variable extraction in property-casualty insurance.

q-fin.MF↗

arXiv2023

Generalization of Graph-Based Active Learning Relaxation Strategies Across Materials

Although density functional theory (DFT) has aided in accelerating the discovery of new materials, such calculations are computationally expensive, especially for high-throughput efforts. This has prompted an explosion in exploration of machine learning assisted techniques to improve the computational efficiency of DFT. In this study, we present a comprehensive investigation of the broader application of Finetuna, an active learning framework to accelerate structural relaxation in DFT with prior information from Open Catalyst Project pretrained graph neural networks. We explore the challenges associated with out-of-domain systems: alcohol ($C_{>2}$) on metal surfaces as larger adsorbates, metal-oxides with spin polarization, and three-dimensional (3D) structures like zeolites and metal-organic-frameworks. By pre-training machine learning models on large datasets and fine-tuning the model along the simulation, we demonstrate the framework's ability to conduct relaxations with fewer DFT calculations. Depending on the similarity of the test systems to the training systems, a more conservative querying strategy is applied. Our best-performing Finetuna strategy reduces the number of DFT single-point calculations by 80% for alcohols and 3D structures, and 42% for oxide systems.

cond-mat.mtrl-sci↗

arXiv2023

Rational design of nanoscale stabilized oxide catalysts for OER with OC22

The efficiency of $H_2$ production via water electrolysis is typically limited to the sluggish oxygen evolution reaction (OER). As such, significant emphasis has been placed upon improving the rate of OER through the anode catalyst. More recently, the Open Catalyst 2022 (OC22) framework has provided a large dataset of density functional theory (DFT) calculations for OER intermediates on the surfaces of oxides. When coupled with state-of-the-art graph neural network models, total energy predictions can be achieved with a mean absolute error as low as 0.22 eV. In this work, we interpolated a database of the total energy predictions for all slabs and OER surface intermediates for 4,119 oxide materials in the original OC22 dataset using pre-trained models from the OC22 framework. This database includes all terminations of all facets up to a maximum Miller index of 1. To demonstrate the full utility of this database, we constructed a flexible screening framework to identify viable candidate anode catalysts for OER under varying reaction conditions for bulk, surface, and nanoscale Pourbaix stability as well as material cost, overpotential, and metastability. From our assessment, we were able to identify 122 and 68 viable candidates for OER under the bulk and nanoscale regime respectively.

cond-mat.mtrl-sci↗

arXiv2022

The Open Catalyst 2022 (OC22) Dataset and Challenges for Oxide Electrocatalysts

The development of machine learning models for electrocatalysts requires a broad set of training data to enable their use across a wide variety of materials. One class of materials that currently lacks sufficient training data is oxides, which are critical for the development of OER catalysts. To address this, we developed the OC22 dataset, consisting of 62,331 DFT relaxations (~9,854,504 single point calculations) across a range of oxide materials, coverages, and adsorbates. We define generalized total energy tasks that enable property prediction beyond adsorption energies; we test baseline performance of several graph neural networks; and we provide pre-defined dataset splits to establish clear benchmarks for future efforts. In the most general task, GemNet-OC sees a ~36% improvement in energy predictions when combining the chemically dissimilar OC20 and OC22 datasets via fine-tuning. Similarly, we achieved a ~19% improvement in total energy predictions on OC20 and a ~9% improvement in force predictions in OC22 when using joint training. We demonstrate the practical utility of a top performing model by capturing literature adsorption energies and important OER scaling relationships. We expect OC22 to provide an important benchmark for models seeking to incorporate intricate long-range electrostatic and magnetic interactions in oxide surfaces. Dataset and baseline models are open sourced, and a public leaderboard is available to encourage continued community developments on the total energy tasks and data.

cond-mat.mtrl-sci↗

arXiv2021

Proton distribution visualization in perovskite nickelate devices utilizing nanofocused X-rays

We use a 30-nm x-ray beam to study the spatially resolved properties of a SmNiO$_3$-based nanodevice that is doped with protons. The x-ray absorption spectra supported by density-functional theory (DFT) simulations show partial reduction of nickel valence in the region with high proton concentration, which leads to the insulating behavior. Concurrently, x-ray diffraction reveals only a small lattice distortion in the doped regions. Together, our results directly show that the knob which proton doping modifies is the electronic valency, and not the crystal lattice. The studies are relevant to on-going efforts to disentangle structural and electronic effects across metal-insulator phase transitions in correlated oxides.

cond-mat.mtrl-sci↗

arXiv2020

The Breakdown of Mott Physics at VO$_2$ Surfaces

Transition metal oxides such as vanadium dioxide (VO$_2$), niobium dioxide (NbO$_2$), and titanium sesquioxide (Ti$_2$O$_3$) are known to undergo a temperature-dependent metal-insulator transition (MIT) in conjunction with a structural transition within their bulk. However, it is not typically discussed how breaking crystal symmetry via surface termination affects the complicated MIT physics. Using synchrotron-based x-ray spectroscopy, low energy electron diffraction (LEED), low energy electron microscopy (LEEM), transmission electron microscopy (TEM), and several other experimental techniques, we show that suppression of the bulk structural transition is a common feature at VO$_2$ surfaces. Our density functional theory (DFT) calculations further suggest that this is due to inherent reconstructions necessary to stabilize the surface, which deviate the electronic structure away from the bulk d$^1$ configuration. Our findings have broader ramifications not only for the characterization of other "Mott-like" MITs, but also for any potential device applications of such materials.

cond-mat.str-el↗

arXiv2020

SAD: Saliency-based Defenses Against Adversarial Examples

With the rise in popularity of machine and deep learning models, there is an increased focus on their vulnerability to malicious inputs. These adversarial examples drift model predictions away from the original intent of the network and are a growing concern in practical security. In order to combat these attacks, neural networks can leverage traditional image processing approaches or state-of-the-art defensive models to reduce perturbations in the data. Defensive approaches that take a global approach to noise reduction are effective against adversarial attacks, however their lossy approach often distorts important data within the image. In this work, we propose a visual saliency based approach to cleaning data affected by an adversarial attack. Our model leverages the salient regions of an adversarial image in order to provide a targeted countermeasure while comparatively reducing loss within the cleaned images. We measure the accuracy of our model by evaluating the effectiveness of state-of-the-art saliency methods prior to attack, under attack, and after application of cleaning methods. We demonstrate the effectiveness of our proposed approach in comparison with related defenses and against established adversarial attack methods, across two saliency datasets. Our targeted approach shows significant improvements in a range of standard statistical and distance saliency metrics, in comparison with both traditional and state-of-the-art approaches.

cs.CV↗

arXiv2019

Grain Boundary Properties of Elemental Metals

The structure and energy of grain boundaries (GBs) are essential for predicting the properties of polycrystalline materials. In this work, we use high-throughput density functional theory calculations workflow to construct the Grain Boundary Database (GBDB), the largest database of DFT-computed grain boundary properties to date. The database currently encompasses 327 GBs of 58 elemental metals, including 10 common twist or symmetric tilt GBs for body-centered cubic (bcc) and face-centered cubic (fcc) systems and the $Σ$7 [0001] twist GB for hexagonal close-packed (hcp) systems. In particular, we demonstrate a novel scaled-structural template approach for HT GB calculations, which reduces the computational cost of converging GB structures by a factor of $\sim 3-6$. The grain boundary energies and work of separation are rigorously validated against previous experimental and computational data. Using this large GB dataset, we develop an improved predictive model for the GB energy of different elements based on the cohesive energy and shear modulus. The open GBDB represent a significant step forward in the availability of first principles GB properties, which we believe would help guide the future design of polycrystalline materials.

cond-mat.mtrl-sci↗

arXiv2019

Anisotropic work function of elemental crystals

The work function is a fundamental electronic property of a solid that varies with the facets of a crystalline surface. It is a crucial parameter in spectroscopy as well as materials design, especially for technologies such as thermionic electron guns and Schottky barriers. In this work, we present the largest database of calculated work functions for elemental crystals to date. This database contains the anisotropic work functions of more than 100 polymorphs of about 72 elements and up to a maximum Miller index of two and three for non-cubic and cubic crystals, respectively. The database has been rigorously validated against previous experimental and computational data where available. We also propose a weighted work function based on the Wulff shape that can be compared to measurements from polycrystalline specimens, and show that this weighted work function can be modeled empirically using simple atomic parameters. Furthermore, for the first time, we were able to analyze simple bond breaking rules for metallic systems beyond a maximum Miller index of one, allowing for a more generalized investigation of work function anisotropy.

cond-mat.mtrl-sci↗

arXiv2017

Accurate Force Field for Molybdenum by Machine Learning Large Materials Data

In this work, we present a highly accurate spectral neighbor analysis potential (SNAP) model for molybdenum (Mo) developed through the rigorous application of machine learning techniques on large materials data sets. Despite Mo's importance as a structural metal, existing force fields for Mo based on the embedded atom and modified embedded atom methods still do not provide satisfactory accuracy on many properties. We will show that by fitting to the energies, forces and stress tensors of a large density functional theory (DFT)-computed dataset on a diverse set of Mo structures, a Mo SNAP model can be developed that achieves close to DFT accuracy in the prediction of a broad range of properties, including energies, forces, stresses, elastic constants, melting point, phonon spectra, surface energies, grain boundary energies, etc. We will outline a systematic model development process, which includes a rigorous approach to structural selection based on principal component analysis, as well as a differential evolution algorithm for optimizing the hyperparameters in the model fitting so that both the model error and the property prediction error can be simultaneously lowered. We expect that this newly developed Mo SNAP model will find broad applications in large-scale, long-time scale simulations.

physics.comp-ph↗