arXiv subjects

Huan He

Publications and source records attributed to Huan He.

At least 37 records · Page 2Linked to original sources

arXiv2025

Benchmarking large language models for biomedical natural language processing applications and recommendations

The rapid growth of biomedical literature poses challenges for manual knowledge curation and synthesis. Biomedical Natural Language Processing (BioNLP) automates the process. While Large Language Models (LLMs) have shown promise in general domains, their effectiveness in BioNLP tasks remains unclear due to limited benchmarks and practical guidelines. We perform a systematic evaluation of four LLMs, GPT and LLaMA representatives on 12 BioNLP benchmarks across six applications. We compare their zero-shot, few-shot, and fine-tuning performance with traditional fine-tuning of BERT or BART models. We examine inconsistencies, missing information, hallucinations, and perform cost analysis. Here we show that traditional fine-tuning outperforms zero or few shot LLMs in most tasks. However, closed-source LLMs like GPT-4 excel in reasoning-related tasks such as medical question answering. Open source LLMs still require fine-tuning to close performance gaps. We find issues like missing information and hallucinations in LLM outputs. These results offer practical insights for applying LLMs in BioNLP.

cs.CL↗

arXiv2025

Non-uniform wave momentum bandgap in biaxial anisotropic photonic time crystals

Photonic time crystals (PTCs) host momentum bandgaps enabling intriguing non-resonant light amplification in propagating waves, but opening substantial bandgaps demands refractive index changes too extreme for conventional nonlinear optics. Here, we introduce momentum bandgaps for non-uniform waves, including evanescent and ghost types, by extending PTCs to biaxial anisotropic photonic time crystals that periodically alternate between uniform biaxial anisotropy and isotropic media over time. We show that ghost waves, unlike evanescent waves, sustain only momentum bandgaps, opening wide bandgaps at even the smallest modulation depths. Moreover, we demonstrate momentum bandgap effects on non-uniform waves that can be amplified, or through decaying modes, selectively attenuated. We find that ghost wave momentum bandgaps uniquely boost refracted over reflected waves under one-way incidence, in stark contrast to balanced amplification seen in both propagating and evanescent waves. Our approach expands time-varying metamaterials by integrating wave characteristics, bridging the gap between conventional nonlinear optics and PTC momentum bandgaps, and shedding new light on extreme manipulation of surface polaritons.

physics.optics↗

arXiv2024

CDEMapper: Enhancing NIH Common Data Element Normalization using Large Language Models

Common Data Elements (CDEs) standardize data collection and sharing across studies, enhancing data interoperability and improving research reproducibility. However, implementing CDEs presents challenges due to the broad range and variety of data elements. This study aims to develop an effective and efficient mapping tool to bridge the gap between local data elements and National Institutes of Health (NIH) CDEs. We propose CDEMapper, a large language model (LLM) powered mapping tool designed to assist in mapping local data elements to NIH CDEs. CDEMapper has three core modules: (1) CDE indexing and embeddings. NIH CDEs were indexed and embedded to support semantic search; (2) CDE recommendations. The tool combines Elasticsearch (BM25 similarity methods) with state of the art GPT services to recommend candidate CDEs and their permissible values; and (3) Human review. Users review and select the NIH CDEs and values that best match their data elements and value sets. We evaluate the tool recommendation accuracy against manually annotated mapping results. CDEMapper offers a publicly available, LLM-powered, and intuitive user interface that consolidates essential and advanced mapping services into a streamlined pipeline. It provides a step by step, quality assured mapping workflow designed with a user-centered approach. The evaluation results demonstrated that augmenting BM25 with GPT embeddings and a ranker consistently enhances CDEMapper mapping accuracy in three different mapping settings across four evaluation datasets. This work opens up the potential of using LLMs to assist with CDE recommendation and human curation when aligning local data elements with NIH CDEs. Additionally, this effort enhances clinical research data interoperability and helps researchers better understand the gaps between local data elements and NIH CDEs.

cs.IR↗

arXiv2024

Me LLaMA: Foundation Large Language Models for Medical Applications

Recent advancements in large language models (LLMs) like ChatGPT and LLaMA show promise in medical applications, yet challenges remain in medical language comprehension. This study presents Me-LLaMA, a new medical LLM family based on open-source LLaMA models, optimized for medical text analysis and diagnosis by leveraging large-scale, domain-specific datasets. The Me-LLaMA family, including foundation models Me-LLaMA 13/70B and their chat-enhanced versions, was developed through continued pre-training and instruction tuning with 129B tokens and 214K samples from biomedical and clinical sources. Training the 70B models required over 100,000 A100 GPU hours. Me-LLaMA's performance was evaluated across six medical text analysis tasks using 12 benchmark datasets and complex clinical case diagnosis, with automatic and human evaluations. Results indicate Me-LLaMA outperforms LLaMA and other open-source medical LLMs in zero-shot and supervised settings. Task-specific tuning further boosts performance, surpassing ChatGPT on 7 of 8 datasets and GPT-4 on 5 of 8. For complex clinical cases, Me-LLaMA achieves performance comparable to ChatGPT and GPT-4. This work underscores the importance of domain-specific data in developing medical LLMs and addresses the high computational costs involved in training, highlighting a balance between pre-training and fine-tuning strategies. Me-LLaMA models are now accessible under user agreements, providing a valuable resource for advancing medical AI.

cs.CL↗

arXiv2024

Obstacle-Aware Length-Matching Routing for Any-Direction Traces in Printed Circuit Board

Emerging applications in Printed Circuit Board (PCB) routing impose new challenges on automatic length matching, including adaptability for any-direction traces with their original routing preserved for interactiveness. The challenges can be addressed through two orthogonal stages: assign non-overlapping routing regions to each trace and meander the traces within their regions to reach the target length. In this paper, mainly focusing on the meandering stage, we propose an obstacle-aware detailed routing approach to optimize the utilization of available space and achieve length matching while maintaining the original routing of traces. Furthermore, our approach incorporating the proposed Multi-Scale Dynamic Time Warping (MSDTW) method can also handle differential pairs against common decoupled problems. Experimental results demonstrate that our approach has effective length-matching routing ability and compares favorably to previous approaches under more complicated constraints.

cs.AR↗

arXiv2024

Anderson Acceleration with Truncated Gram-Schmidt

Anderson Acceleration (AA) is a popular algorithm designed to enhance the convergence of fixed-point iterations. In this paper, we introduce a variant of AA based on a Truncated Gram-Schmidt process (AATGS) which has a few advantages over the classical AA. In particular, an attractive feature of AATGS is that its iterates obey a three-term recurrence in the situation when it is applied to solving symmetric linear problems and this can lead to a considerable reduction of memory and computational costs. We analyze the convergence of AATGS in both full-depth and limited-depth scenarios and establish its equivalence to the classical AA in the linear case. We also report on the effectiveness of AATGS through a set of numerical experiments, ranging from solving nonlinear partial differential equations to tackling nonlinear optimization problems. In particular, the performance of the method is compared with that of the classical AA algorithms.

math.NA↗

arXiv2024

Ultrahigh Stability of O-Sublattice in $β$-Ga$_2$O$_3$

Recently reported remarkably high radiation tolerance of $γ$/$β$-Ga$_2$O$_3$ double-polymorphic structure brings this ultrawide bandgap semiconductor to the frontiers of power electronics applications that are able to operate in challenging environments. Understanding the mechanism of radiation tolerance is crucial for further material modification and tailoring of the desired properties. In this study, we employ machine-learning-enhanced atomistic simulations to assess the stability of both the gallium (Ga) and oxygen (O) sublattices under various levels of damage. Our study uncovers the remarkable resilience and stability of the O-sublattice, attributing this property to the strong tendency of recovery of the O defects, especially within the stronger disordered regions. Interestingly, we observe the opposite behavior of the Ga defects that display enhanced stability in the same regions of increased disorder. Moreover, we observe that highly defective $β$-Ga$_2$O$_3$ is able to transform into $γ$-Ga$_2$O$_3$ upon annealing due to preserved lattice organization of the O-sublattice. This result clearly manifests that the ultrahigh stability of the O-sublattice provides the backbone for the exceptional radiation tolerance of the $γ$/$β$ double-polymorphic structure. These computational insights closely align with experimental observations, opening avenues for further exploration of polymorphism in Ga$_2$O$_3$ and potentially in analogous polymorphic families spanning a broad range of diverse materials of complex polymorphic nature.

cond-mat.mtrl-sci↗

arXiv2024

NLTGCR: A class of Nonlinear Acceleration Procedures based on Conjugate Residuals

This paper develops a new class of nonlinear acceleration algorithms based on extending conjugate residual-type procedures from linear to nonlinear equations. The main algorithm has strong similarities with Anderson acceleration as well as with inexact Newton methods - depending on which variant is implemented. We prove theoretically and verify experimentally, on a variety of problems from simulation experiments to deep learning applications, that our method is a powerful accelerated iterative algorithm.

math.NA↗

arXiv2024

Threshold displacement energy map of Frenkel pair generation in $\rm Ga_2O_3$ from machine-learning-driven molecular dynamics simulations

$β$ phase gallium oxide ($β$-$\rm Ga_2O_3$) demonstrates tremendous potential for electronics applications and offers promising prospects for integration into future space systems with the necessity of high radiation resistance. Therefore, a comprehensive understanding of the threshold displacement energy (TDE) and the radiation-induced formation of Frenkel pairs (FPs) in this material is vital but has not yet been thoroughly studied. In this work, we performed over 5,000 molecular dynamics simulations using our machine-learning potentials to determine the TDE and investigate the formation of FPs. The average TDEs for the two Ga sites, Ga1 (tetrahedral site) and Ga2 (octahedral site), are 22.9 and 20.0 eV, respectively. While the average TDEs for the three O sites are nearly uniform, ranging from 17.0 to 17.4 eV. The generated TDE maps reveal significant differences in displacement behavior between these five atomic sites. Our developed defect identification methods successfully categorize various types of FPs in this material, with more than ten types of Ga FPs being produced during our simulations. O atoms are found to form two main types of FPs and the O split interstitial site on O1 site is most common. Finally, the recombination behavior and barriers of Ga and O FPs indicate that the O FP has a higher possibility of recovery upon annealing. Our findings provide important insights into the studies of radiation damage and defects in $\rm Ga_2O_3$ and can contribute to the design and development of $\rm Ga_2O_3$-based devices

cond-mat.mtrl-sci↗

arXiv2023

Encoding Time-Series Explanations through Self-Supervised Model Behavior Consistency

Interpreting time series models is uniquely challenging because it requires identifying both the location of time series signals that drive model predictions and their matching to an interpretable temporal pattern. While explainers from other modalities can be applied to time series, their inductive biases do not transfer well to the inherently challenging interpretation of time series. We present TimeX, a time series consistency model for training explainers. TimeX trains an interpretable surrogate to mimic the behavior of a pretrained time series model. It addresses the issue of model faithfulness by introducing model behavior consistency, a novel formulation that preserves relations in the latent space induced by the pretrained model with relations in the latent space induced by TimeX. TimeX provides discrete attribution maps and, unlike existing interpretability methods, it learns a latent space of explanations that can be used in various ways, such as to provide landmarks to visually aggregate similar explanations and easily recognize temporal patterns. We evaluate TimeX on eight synthetic and real-world datasets and compare its performance against state-of-the-art interpretability methods. We also conduct case studies using physiological time series. Quantitative evaluations demonstrate that TimeX achieves the highest or second-highest performance in every metric compared to baselines across all datasets. Through case studies, we show that the novel components of TimeX show potential for training faithful, interpretable models that capture the behavior of pretrained time series models.

cs.LG↗

arXiv2023

Universal radiation tolerant semiconductor

Radiation tolerance is determined as the ability of crystalline materials to withstand the accumulation of the radiation induced disorder. Nevertheless, for sufficiently high fluences, in all by far known semiconductors it ends up with either very high disorder levels or amorphization. Here we show that gamma/beta double polymorph Ga2O3 structures exhibit remarkably high radiation tolerance. Specifically, for room temperature experiments, they tolerate a disorder equivalent to hundreds of displacements per atom, without severe degradations of crystallinity; in comparison with, e.g., Si amorphizable already with the lattice atoms displaced just once. We explain this behavior by an interesting combination of the Ga- and O- sublattice properties in gamma-Ga2O3. In particular, O-sublattice exhibits a strong recrystallization trend to recover the face-centered-cubic stacking despite the stronger displacement of O atoms compared to Ga during the active periods of cascades. Notably, we also explained the origin of the beta-to-gamma Ga2O3 transformation, as a function of the increased disorder in beta-Ga2O3 and studied the phenomena as a function of the chemical nature of the implanted atoms. As a result, we conclude that gamma/beta double polymorph Ga2O3 structures, in terms of their radiation tolerance properties, benchmark a class of universal radiation tolerant semiconductors.

cond-mat.mtrl-sci↗

arXiv2023

Reducing operator complexity in Algebraic Multigrid with Machine Learning Approaches

We propose a data-driven and machine-learning-based approach to compute non-Galerkin coarse-grid operators in algebraic multigrid (AMG) methods, addressing the well-known issue of increasing operator complexity. Guided by the AMG theory on spectrally equivalent coarse-grid operators, we have developed novel ML algorithms that utilize neural networks (NNs) combined with smooth test vectors from multigrid eigenvalue problems. The proposed method demonstrates promise in reducing the complexity of coarse-grid operators while maintaining overall AMG convergence for solving parametric partial differential equation (PDE) problems. Numerical experiments on anisotropic rotated Laplacian and linear elasticity problems are provided to showcase the performance and compare with existing methods for computing non-Galerkin coarse-grid operators.

math.NA↗

arXiv2023

Domain Adaptation for Time Series Under Feature and Label Shifts

Unsupervised domain adaptation (UDA) enables the transfer of models trained on source domains to unlabeled target domains. However, transferring complex time series models presents challenges due to the dynamic temporal structure variations across domains. This leads to feature shifts in the time and frequency representations. Additionally, the label distributions of tasks in the source and target domains can differ significantly, posing difficulties in addressing label shifts and recognizing labels unique to the target domain. Effectively transferring complex time series models remains a formidable problem. We present Raincoat, the first model for both closed-set and universal domain adaptation on complex time series. Raincoat addresses feature and label shifts by considering both temporal and frequency features, aligning them across domains, and correcting for misalignments to facilitate the detection of private labels. Additionally, Raincoat improves transferability by identifying label shifts in target domains. Our experiments with 5 datasets and 13 state-of-the-art UDA methods demonstrate that Raincoat can improve transfer learning performance by up to 16.33% and can handle both closed-set and universal domain adaptation.

cs.LG↗

arXiv2023

Complex $\mathrm{Ga}_{2}\mathrm{O}_{3}$ Polymorphs Explored by Accurate and General-Purpose Machine-Learning Interatomic Potentials

$\mathrm{Ga}_{2}\mathrm{O}_{3}$ is a wide-bandgap semiconductor of emergent importance for applications in electronics and optoelectronics. However, vital information of the properties of complex coexisting $\mathrm{Ga}_{2}\mathrm{O}_{3}$ polymorphs and low-symmetry disordered structures is missing. In this work, we develop two types of kernel-based machine-learning Gaussian approximation potentials (ML-GAPs) for $\mathrm{Ga}_{2}\mathrm{O}_{3}$ with high accuracy for $β$/$κ$/$α$/$δ$/$γ$ polymorphs and generality for disordered stoichiometric structures. We release two versions of interatomic potentials in parallel, namely soapGAP and tabGAP, for excellent accuracy and exceeding speedup, respectively. We systematically show that both the soapGAP and tabGAP can reproduce the structural properties of all the five polymorphs in an exceptional agreement with ab initio results, meanwhile boost the computational efficiency with $5\times10^{2}$ and $2\times10^{5}$ computing speed increases compared to density functional theory, respectively. The results show that the liquid-solid phase transition proceeds in three different stages, a "slow transition", "fast transition" and "only Ga migration". We show that this complex dynamics can be understood in terms of different behavior of O and Ga sublattices in the interfacial layer.

cond-mat.mtrl-sci↗

arXiv2023

GNNDelete: A General Strategy for Unlearning in Graph Neural Networks

Graph unlearning, which involves deleting graph elements such as nodes, node labels, and relationships from a trained graph neural network (GNN) model, is crucial for real-world applications where data elements may become irrelevant, inaccurate, or privacy-sensitive. However, existing methods for graph unlearning either deteriorate model weights shared across all nodes or fail to effectively delete edges due to their strong dependence on local graph neighborhoods. To address these limitations, we introduce GNNDelete, a novel model-agnostic layer-wise operator that optimizes two critical properties, namely, Deleted Edge Consistency and Neighborhood Influence, for graph unlearning. Deleted Edge Consistency ensures that the influence of deleted elements is removed from both model weights and neighboring representations, while Neighborhood Influence guarantees that the remaining model knowledge is preserved after deletion. GNNDelete updates representations to delete nodes and edges from the model while retaining the rest of the learned knowledge. We conduct experiments on seven real-world graphs, showing that GNNDelete outperforms existing approaches by up to 38.8% (AUC) on edge, node, and node feature deletion tasks, and 32.2% on distinguishing deleted edges from non-deleted ones. Additionally, GNNDelete is efficient, taking 12.3x less time and 9.3x less space than retraining GNN from scratch on WordNet18.

cs.LG↗

arXiv2023

Development of an Extractive Clinical Question Answering Dataset with Multi-Answer and Multi-Focus Questions

Background: Extractive question-answering (EQA) is a useful natural language processing (NLP) application for answering patient-specific questions by locating answers in their clinical notes. Realistic clinical EQA can have multiple answers to a single question and multiple focus points in one question, which are lacking in the existing datasets for development of artificial intelligence solutions. Objective: Create a dataset for developing and evaluating clinical EQA systems that can handle natural multi-answer and multi-focus questions. Methods: We leveraged the annotated relations from the 2018 National NLP Clinical Challenges (n2c2) corpus to generate an EQA dataset. Specifically, the 1-to-N, M-to-1, and M-to-N drug-reason relations were included to form the multi-answer and multi-focus QA entries, which represent more complex and natural challenges in addition to the basic one-drug-one-reason cases. A baseline solution was developed and tested on the dataset. Results: The derived RxWhyQA dataset contains 96,939 QA entries. Among the answerable questions, 25% require multiple answers, and 2% ask about multiple drugs within one question. There are frequent cues observed around the answers in the text, and 90% of the drug and reason terms occur within the same or an adjacent sentence. The baseline EQA solution achieved a best f1-measure of 0.72 on the entire dataset, and on specific subsets, it was: 0.93 on the unanswerable questions, 0.48 on single-drug questions versus 0.60 on multi-drug questions, 0.54 on the single-answer questions versus 0.43 on multi-answer questions. Discussion: The RxWhyQA dataset can be used to train and evaluate systems that need to handle multi-answer and multi-focus questions. Specifically, multi-answer EQA appears to be challenging and therefore warrants more investment in research.

cs.CL↗

arXiv2023

MedDiff: Generating Electronic Health Records using Accelerated Denoising Diffusion Model

Due to patient privacy protection concerns, machine learning research in healthcare has been undeniably slower and limited than in other application domains. High-quality, realistic, synthetic electronic health records (EHRs) can be leveraged to accelerate methodological developments for research purposes while mitigating privacy concerns associated with data sharing. The current state-of-the-art model for synthetic EHR generation is generative adversarial networks, which are notoriously difficult to train and can suffer from mode collapse. Denoising Diffusion Probabilistic Models, a class of generative models inspired by statistical thermodynamics, have recently been shown to generate high-quality synthetic samples in certain domains. It is unknown whether these can generalize to generation of large-scale, high-dimensional EHRs. In this paper, we present a novel generative model based on diffusion models that is the first successful application on electronic health records. Our model proposes a mechanism to perform class-conditional sampling to preserve label information. We also introduce a new sampling strategy to accelerate the inference speed. We empirically show that our model outperforms existing state-of-the-art synthetic EHR generation methods.

cs.LG↗

arXiv2023

Stationary Charge Radiation in Anisotropic Photonic Time Crystals

Time metamaterials exhibit a great potential for wave manipulation, drawing increasing attention in recent years. Here, we explore the exotic wave dynamics in an anisotropic photonic time crystal (APTC), formed by an anisotropic medium whose optical properties are uniformly and periodically changed in time. Based on a temporal transfer matrix formalism, we show that a stationary charge embedded in an APTC can emit radiation, in contrast to the case of an isotropic photonic time crystal, and its distribution in momentum space is controlled by the APTC band structure. Our approach extends the functionalities of time metamaterials, offering new opportunities for simultaneous radiation generation and control, with implications for both classical and quantum applications.

physics.optics↗