Elias Ioup

Elias Ioup

Auto-generated profile, based on publicly available information

Elias Ioup

US Naval Research Laboratory

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 January 2026 in ACM Computing Surveys
Verified
Journal Article
Open access
Md Meftahul FerdausMd Meftahul Ferdaus
Md Meftahul Ferdaus
MAMahdi Abdelguerfi
Mahdi Abdelguerfi
ELElias Loup
Elias Loup
KNKendall N. Niles
Kendall N. Niles
KPKen Pathak
Ken Pathak
SSSteven Sloan
Steven Sloan
Large Language Models (LLMs) are advancing rapidly and promising transformation across fields but pose challenges in oversight, ethics, and user trust. This review addresses trust issues like unintentional harms, opacity, vulnerability, misalignment with values, and environmental impact, all of which affect trust. Factors undermining trust include societal biases, opaque processes, misuse potential, and technology evolution challenges, especially in finance, healthcare, education, and policy. Recommended solutions include ethical oversight, industry accountability, regulation, and public involvement to reshape AI norms and incorporate ethics into development. A framework assesses trust in LLMs, analyzing trust dynamics and providing guidelines for responsible AI development. The review highlights limitations in building trustworthy AI, aiming to create a transparent and accountable ecosystem that maximizes benefits and minimizes risks, offering guidance for researchers, policymakers, and industry in fostering trust and ensuring responsible use of LLMs. We validate our frameworks through comprehensive experimental assessment across seven contemporary models, demonstrating substantial improvements in trustworthiness characteristics and identifying important disagreements with existing literature. Both theoretical foundations and empirical validation are provided in comprehensive supplementary materials.

14 May 2024 in IEEE Journal of Oceanic Engineering
Verified
Journal Article
Closed access
Austin SchmidtAustin Schmidt
Austin Schmidt
PPPujan Pokhrel
Pujan Pokhrel
MAMahdi Abdelguerfi
Mahdi Abdelguerfi
EIElias Ioup
Elias Ioup
DDDavid Dobson
David Dobson
Methodologies inspired by physics-informed neural networks (PINNs) were used to forecast observations recorded by stationary ocean buoys. We combined buoy observations with numerical models to train surrogate deep learning networks that performed better than with either data alone. Numerical model outputs were collected from two sources for training and regularization: the hybrid circulation ocean model and the fifth European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis experiment. A hyperparameter determines the ratio of observational and modeled data to be used in the training procedure, so we conducted a grid search to find the most performant ratio. Overall, the technique improved the general forecast performance compared with nonregularized models. Under specific circumstances, the regularization mechanism enabled the PINN models to be more accurate than the numerical models. This demonstrates the utility of combining various climate models and sensor observations to improve surrogate modeling.

8 October 2024
Verified
Conference Paper
Open access
Md Meftahul FerdausMd Meftahul Ferdaus
Md Meftahul Ferdaus
MAMahdi Abdelguerfi
Mahdi Abdelguerfi
EIElias Ioup
Elias Ioup
DDDavid Dobson
David Dobson
KNKendall N. Niles
Kendall N. Niles
SSSteven Sloan
Steven Sloan

10 March 2026 in ACM Computing Surveys
Verified
Journal Article
Open access
Md Meftahul FerdausMd Meftahul Ferdaus
Md Meftahul Ferdaus
KNKendall N. Niles
Kendall N. Niles
JTJoe Tom
Joe Tom
MAMahdi Abdelguerfi
Mahdi Abdelguerfi
EIElias Ioup
Elias Ioup
Few-shot learning (FSL) and data-efficient learning paradigms enable object detection models to recognize novel classes from minimally annotated examples, addressing expensive data-labeling challenges. This systematic survey examines recent advances in few-shot, semi-supervised, sparsely-supervised, and weakly-supervised approaches for video and 3D object detection, focusing on developments through foundation models and vision-language model integration. For video object detection, techniques including tube proposals, temporal matching networks, motion-guided approaches, and temporal consistency-based semi-supervised methods utilize spatiotemporal relationships for efficient novel class adaptation, with recent architectures achieving substantial gains from 33 to 48 average precision in few-shot scenarios. For 3D object detection, specialized approaches address point cloud sparsity and texture limitations through uncertainty-aware methods, geometric learning, and multimodal fusion, with sparsely-supervised techniques achieving competitive performance using only 2% of annotations, enabling practical deployment in autonomous driving and robotics. The survey analyzes methodological advances including meta-learning, transfer learning, pseudo-label generation, contrastive instance mining, and foundation model integration across applications spanning autonomous driving, surveillance, robotics, industrial control, and medical imaging. By examining developments across multiple supervision paradigms, this work highlights data-efficient learning’s potential for minimizing annotation requirements and enabling robust real-world deployment across temporal, spatial, and multimodal domains.

5 February 2015 in IEEE Internet Computing
Verified
Journal Article
Closed access
EIElias Ioup
Elias Ioup
ZYZhao Yang
Zhao Yang
BBBrent Barré
Brent Barré
JSJohn Sample
John Sample
KSKevin B. Shaw
Kevin B. Shaw
MAMahdi Abdelguerfi
Mahdi Abdelguerfi
The Geography Markup Language (GML) - the existing standard for encoding geospatial data - has no mechanism for annotating such data with uncertainty. To address this issue while supporting the geospatial community's existing data and service standards, the authors extend GML to enable uncertainty markup. They demonstrate this extension's use with some common geospatial data types and Web services. The result is a robust capability to share error information while maintaining compatibility with existing geospatial data clients.