arXiv subjects

Liya Wang

Publications and source records attributed to Liya Wang.

16 recordsLinked to original sources

arXiv2025

Enterprise Large Language Model Evaluation Benchmark

Large Language Models (LLMs) ) have demonstrated promise in boosting productivity across AI-powered tools, yet existing benchmarks like Massive Multitask Language Understanding (MMLU) inadequately assess enterprise-specific task complexities. We propose a 14-task framework grounded in Bloom's Taxonomy to holistically evaluate LLM capabilities in enterprise contexts. To address challenges of noisy data and costly annotation, we develop a scalable pipeline combining LLM-as-a-Labeler, LLM-as-a-Judge, and corrective retrieval-augmented generation (CRAG), curating a robust 9,700-sample benchmark. Evaluation of six leading models shows open-source contenders like DeepSeek R1 rival proprietary models in reasoning tasks but lag in judgment-based scenarios, likely due to overthinking. Our benchmark reveals critical enterprise performance gaps and offers actionable insights for model optimization. This work provides enterprises a blueprint for tailored evaluations and advances practical LLM deployment.

cs.AI↗

arXiv2023

Adapting Sentence Transformers for the Aviation Domain

Learning effective sentence representations is crucial for many Natural Language Processing (NLP) tasks, including semantic search, semantic textual similarity (STS), and clustering. While multiple transformer models have been developed for sentence embedding learning, these models may not perform optimally when dealing with specialized domains like aviation, which has unique characteristics such as technical jargon, abbreviations, and unconventional grammar. Furthermore, the absence of labeled datasets makes it difficult to train models specifically for the aviation domain. To address these challenges, we propose a novel approach for adapting sentence transformers for the aviation domain. Our method is a two-stage process consisting of pre-training followed by fine-tuning. During pre-training, we use Transformers and Sequential Denoising AutoEncoder (TSDAE) with aviation text data as input to improve the initial model performance. Subsequently, we fine-tune our models using a Natural Language Inference (NLI) dataset in the Sentence Bidirectional Encoder Representations from Transformers (SBERT) architecture to mitigate overfitting issues. Experimental results on several downstream tasks show that our adapted sentence transformers significantly outperform general-purpose transformers, demonstrating the effectiveness of our approach in capturing the nuances of the aviation domain. Overall, our work highlights the importance of domain-specific adaptation in developing high-quality NLP solutions for specialized industries like aviation.

cs.CL↗

arXiv2023

AviationGPT: A Large Language Model for the Aviation Domain

The advent of ChatGPT and GPT-4 has captivated the world with large language models (LLMs), demonstrating exceptional performance in question-answering, summarization, and content generation. The aviation industry is characterized by an abundance of complex, unstructured text data, replete with technical jargon and specialized terminology. Moreover, labeled data for model building are scarce in this domain, resulting in low usage of aviation text data. The emergence of LLMs presents an opportunity to transform this situation, but there is a lack of LLMs specifically designed for the aviation domain. To address this gap, we propose AviationGPT, which is built on open-source LLaMA-2 and Mistral architectures and continuously trained on a wealth of carefully curated aviation datasets. Experimental results reveal that AviationGPT offers users multiple advantages, including the versatility to tackle diverse natural language processing (NLP) problems (e.g., question-answering, summarization, document writing, information extraction, report querying, data cleaning, and interactive data exploration). It also provides accurate and contextually relevant responses within the aviation domain and significantly improves performance (e.g., over a 40% performance gain in tested cases). With AviationGPT, the aviation industry is better equipped to address more complex research problems and enhance the efficiency and safety of National Airspace System (NAS) operations.

cs.CL↗

arXiv2023

Remote Sensing Scene Classification with Masked Image Modeling (MIM)

Remote sensing scene classification has been extensively studied for its critical roles in geological survey, oil exploration, traffic management, earthquake prediction, wildfire monitoring, and intelligence monitoring. In the past, the Machine Learning (ML) methods for performing the task mainly used the backbones pretrained in the manner of supervised learning (SL). As Masked Image Modeling (MIM), a self-supervised learning (SSL) technique, has been shown as a better way for learning visual feature representation, it presents a new opportunity for improving ML performance on the scene classification task. This research aims to explore the potential of MIM pretrained backbones on four well-known classification datasets: Merced, AID, NWPU-RESISC45, and Optimal-31. Compared to the published benchmarks, we show that the MIM pretrained Vision Transformer (ViTs) backbones outperform other alternatives (up to 18% on top 1 accuracy) and that the MIM technique can learn better feature representation than the supervised learning counterparts (up to 5% on top 1 accuracy). Moreover, we show that the general-purpose MIM-pretrained ViTs can achieve competitive performance as the specially designed yet complicated Transformer for Remote Sensing (TRS) framework. Our experiment results also provide a performance baseline for future studies.

cs.CV↗

arXiv2023

Aerial Image Object Detection With Vision Transformer Detector (ViTDet)

The past few years have seen an increased interest in aerial image object detection due to its critical value to large-scale geo-scientific research like environmental studies, urban planning, and intelligence monitoring. However, the task is very challenging due to the birds-eye view perspective, complex backgrounds, large and various image sizes, different appearances of objects, and the scarcity of well-annotated datasets. Recent advances in computer vision have shown promise tackling the challenge. Specifically, Vision Transformer Detector (ViTDet) was proposed to extract multi-scale features for object detection. The empirical study shows that ViTDet's simple design achieves good performance on natural scene images and can be easily embedded into any detector architecture. To date, ViTDet's potential benefit to challenging aerial image object detection has not been explored. Therefore, in our study, 25 experiments were carried out to evaluate the effectiveness of ViTDet for aerial image object detection on three well-known datasets: Airbus Aircraft, RarePlanes, and Dataset of Object DeTection in Aerial images (DOTA). Our results show that ViTDet can consistently outperform its convolutional neural network counterparts on horizontal bounding box (HBB) object detection by a large margin (up to 17% on average precision) and that it achieves the competitive performance for oriented bounding box (OBB) object detection. Our results also establish a baseline for future research.

cs.CV↗

arXiv2021

Deep Learning for Flight Demand Forecasting

Inspired by the success of deep learning (DL) in natural language processing (NLP), we applied cutting-edge DL techniques to predict flight departure demand in a strategic time horizon (4 hours or longer). This work was conducted in support of a MITRE-developed mobile application, Pacer, which displays predicted departure demand to general aviation (GA) flight operators so they can have better situation awareness of the potential for departure delays during busy periods. Field demonstrations involving Pacer's previously designed rule-based prediction method showed that the prediction accuracy of departure demand still has room for improvement. This research strives to improve prediction accuracy from two key aspects: better data sources and robust forecasting algorithms. We leveraged two data sources, Aviation System Performance Metrics (ASPM) and System Wide Information Management (SWIM), as our input. We then trained forecasting models with DL techniques of sequence to sequence (seq2seq) and seq2seq with attention. The case study has shown that our seq2seq with attention performs best among four forecasting algorithms tested. In addition, with better data sources, seq2seq with attention can reduce mean squared error (mse) over 60%, compared to the classical autoregressive (AR) forecasting method.

cs.LG↗

arXiv2021

Flight Demand Forecasting with Transformers

Transformers have become the de-facto standard in the natural language processing (NLP) field. They have also gained momentum in computer vision and other domains. Transformers can enable artificial intelligence (AI) models to dynamically focus on certain parts of their input and thus reason more effectively. Inspired by the success of transformers, we adopted this technique to predict strategic flight departure demand in multiple horizons. This work was conducted in support of a MITRE-developed mobile application, Pacer, which displays predicted departure demand to general aviation (GA) flight operators so they can have better situation awareness of the potential for departure delays during busy periods. Field demonstrations involving Pacer's previously designed rule-based prediction method showed that the prediction accuracy of departure demand still has room for improvement. This research strives to improve prediction accuracy from two key aspects: better data sources and robust forecasting algorithms. We leveraged two data sources, Aviation System Performance Metrics (ASPM) and System Wide Information Management (SWIM), as our input. We then trained forecasting models with temporal fusion transformer (TFT) for five different airports. Case studies show that TFTs can perform better than traditional forecasting methods by large margins, and they can result in better prediction across diverse airports and with better interpretability.

cs.LG↗

arXiv2021

Multi-Airport Delay Prediction with Transformers

Airport performance prediction with a reasonable look-ahead time is a challenging task and has been attempted by various prior research. Traffic, demand, weather, and traffic management actions are all critical inputs to any prediction model. In this paper, a novel approach based on Temporal Fusion Transformer (TFT) was proposed to predict departure and arrival delays simultaneously for multiple airports at once. This approach can capture complex temporal dynamics of the inputs known at the time of prediction and then forecast selected delay metrics up to four hours into the future. When dealing with weather inputs, a self-supervised learning (SSL) model was developed to encode high-dimensional weather data into a much lower-dimensional representation to make the training of TFT more efficiently and effectively. The initial results show that the TFT-based delay prediction model achieves satisfactory performance measured by smaller prediction errors on a testing dataset. In addition, the interpretability analysis of the model outputs identifies the important input factors for delay prediction. The proposed approach is expected to help air traffic managers or decision makers gain insights about traffic management actions on delay mitigation and once operationalized, provide enough lead time to plan for predicted performance degradation.

cs.LG↗

arXiv2021

Helicopter Track Identification with Autoencoder

Computing power, big data, and advancement of algorithms have led to a renewed interest in artificial intelligence (AI), especially in deep learning (DL). The success of DL largely lies on data representation because different representations can indicate to a degree the different explanatory factors of variation behind the data. In the last few year, the most successful story in DL is supervised learning. However, to apply supervised learning, one challenge is that data labels are expensive to get, noisy, or only partially available. With consideration that we human beings learn in an unsupervised way; self-supervised learning methods have garnered a lot of attention recently. A dominant force in self-supervised learning is the autoencoder, which has multiple uses (e.g., data representation, anomaly detection, denoise). This research explored the application of an autoencoder to learn effective data representation of helicopter flight track data, and then to support helicopter track identification. Our testing results are promising. For example, at Phoenix Deer Valley (DVT) airport, where 70% of recorded flight tracks have missing aircraft types, the autoencoder can help to identify twenty-two times more helicopters than otherwise detectable using rule-based methods; for Grand Canyon West Airport (1G4) airport, the autoencoder can identify thirteen times more helicopters than a current rule-based approach. Our approach can also identify mislabeled aircraft types in the flight track data and find true types for records with pseudo aircraft type labels such as HELO. With improved labelling, studies using these data sets can produce more reliable results.

cs.LG↗

arXiv2020

Autoencoding Features for Aviation Machine Learning Problems

The current practice of manually processing features for high-dimensional and heterogeneous aviation data is labor-intensive, does not scale well to new problems, and is prone to information loss, affecting the effectiveness and maintainability of machine learning (ML) procedures. This research explored an unsupervised learning method, autoencoder, to extract effective features for aviation machine learning problems. The study explored variants of autoencoders with the aim of forcing the learned representations of the input to assume useful properties. A flight track anomaly detection autoencoder was developed to demonstrate the versatility of the technique. The research results show that the autoencoder can not only automatically extract effective features for the flight track data, but also efficiently deep clean data, thereby reducing the workload of data scientists. Moreover, the research leveraged transfer learning to efficiently train models for multiple airports. Transfer learning can reduce model training times from days to hours, as well as improving model performance. The developed applications and techniques are shared with the whole aviation community to improve effectiveness of ongoing and future machine learning studies.

cs.LG↗

arXiv2019

Molecular dynamics simulations of active Brownian particles in dilute suspension: diffusion in free space and distribution in confinement

In this work, we report a new method to simulate active Brownian particles (ABPs) in molecular dynamics (MD) simulations. Immersed in a fluid, each ABP consists of a head particle and a spherical phantom region of fluid where the flagellum of a microswimmer takes effect. The orientation of the active particle is governed by a stochastic dynamics, with the orientational persistence time determined by the rotational diffusivity. To hydrodynamically drive the active particle as a pusher, a pair of active forces are exerted on the head particle and the phantom fluid region respectively. The active velocity measured along the particle orientation is proportional to the magnitude of the active force. The effective diffusion coefficient of the active particle is first measured in free space, showing semi-quantitative agreement with the analytical result predicted by a minimal model for ABPs. We then turn to the probability distribution of the active particle in confinement potential. We find that the stationary particle distribution undergoes an evolution from the Boltzmann-type to non-Boltzmann distribution as the orientational persistence time is increased relative to the relaxation time in the potential well. From the stationary distribution in confinement potential, the active part of the diffusion coefficient is measured and compared to that obtained in free space, showing a good semi-quantitative agreement while the orientational persistence time varies greatly relative to the relaxation time.

cond-mat.soft↗

arXiv2014

CloudSTRUCTURE: infer population STRUCTURE on the cloud

We present CloudSTRUCTURE, an application for running parallel analyses with the population genetics program STRUCTURE. The HPC ready application, powered by iPlant cyber-infrastructure, provides a fast (by parallelization) and convenient (through a user friendly GUI) way to calculate like-lihood values across multiple values of K (number of genetic groups) and numbers of iterations. The results are automati-cally summarized for easier determination of the K value that best fit the data. In addition, CloudSTRUCTURE will reformat STRUCTURE output for use in downstream programs, such as TASSEL for association analysis with population structure ef-fects stratified.

q-bio.QM↗

arXiv2014

The relationships among GC content, nucleosome occupancy, and exon size

The average size of internal translated exons, ranging from 120 to 165 nt across metazoans, is approximately the size of the typical mononucleosome (147 nt). Genome-wide study has also shown that nucleosome occupancy is significantly higher in exons than in introns, which might indicate that the evolution of exon size is related to its nucleosome occupancy. By grouping exons by the GC contents of their flanking introns, we show that the average exon size is positively correlated with its GC content. Using the sequencing data from direct mapping of Homo sapiens nucleosomes with limited nuclease digestion, we show that the level of nucleosome occupancy is also positively correlated with the exon GC content in a similar fashion. We then demonstrated that exon size is positively correlated with their nucleosome occupancy. The strong correlation between exon size and the nucleosome occupancy suggests that chromatin organization may be related to the evolution of exon sizes.

q-bio.GN↗

arXiv2010

Pattern Recognition in Collective Cognitive Systems: Hybrid Human-Machine Learning (HHML) By Heterogeneous Ensembles

The ubiquitous role of the cyber-infrastructures, such as the WWW, provides myriad opportunities for machine learning and its broad spectrum of application domains taking advantage of digital communication. Pattern classification and feature extraction are among the first applications of machine learning that have received extensive attention. The most remarkable achievements have addressed data sets of moderate-to-large size. The 'data deluge' in the last decade or two has posed new challenges for AI researchers to design new, effective and accurate algorithms for similar tasks using ultra-massive data sets and complex (natural or synthetic) dynamical systems. We propose a novel principled approach to feature extraction in hybrid architectures comprised of humans and machines in networked communication, who collaborate to solve a pre-assigned pattern recognition (feature extraction) task. There are two practical considerations addressed below: (1) Human experts, such as plant biologists or astronomers, often use their visual perception and other implicit prior knowledge or expertise without any obvious constraints to search for the significant features, whereas machines are limited to a pre-programmed set of criteria to work with; (2) in a team collaboration of collective problem solving, the human experts have diverse abilities that are complementary, and they learn from each other to succeed in cognitively complex tasks in ways that are still impossible imitate by machines.

cs.AI↗

arXiv2010

Applications of Machine Learning Methods to Quantifying Phenotypic Traits that Distinguish the Wild Type from the Mutant Arabidopsis Thaliana Seedlings during Root Gravitropism

Post-genomic research deals with challenging problems in screening genomes of organisms for particular functions or potential for being the targets of genetic engineering for desirable biological features. 'Phenotyping' of wild type and mutants is a time-consuming and costly effort by many individuals. This article is a preliminary progress report in research on large-scale automation of phenotyping steps (imaging, informatics and data analysis) needed to study plant gene-proteins networks that influence growth and development of plants. Our results undermine the significance of phenotypic traits that are implicit in patterns of dynamics in plant root response to sudden changes of its environmental conditions, such as sudden re-orientation of the root tip against the gravity vector. Including dynamic features besides the common morphological ones has paid off in design of robust and accurate machine learning methods to automate a typical phenotyping scenario, i.e. to distinguish the wild type from the mutants.

q-bio.QM↗

arXiv2010

Parallel Evolutionary Computation in Very Large Scale Eigenvalue Problems

The history of research on eigenvalue problems is rich with many outstanding contributions. Nonetheless, the rapidly increasing size of data sets requires new algorithms for old problems in the context of extremely large matrix dimensions. This paper reports on a new method for finding eigenvalues of very large matrices by a synthesis of evolutionary computation, parallel programming, and empirical stochastic search. The direct design of our method has the added advantage that it could be adapted to extend many algorithmic variants of solutions of generalized eigenvalue problems to improve the accuracy of our algorithms. The preliminary evaluation results are encouraging and demonstrate the method's efficiency and practicality.

cs.DC↗