Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “semantic”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,207 records · Page 67
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Scaling Building Energy Audits through Machine Learning Methods on Novel Drone Image Data

Building energy audits are time-consuming and labor-intensive. This paper describes a new method using machine learning (ML) techniques on novel data sources (drone images) to improve the identification of building characteristics and retrofit opportunities, and thereby reduce the effort for audits. The new ML method includes: (1) Building footprint extraction using line extraction, polygonization, and polygon-merging, (2) Building envelope extraction using PIX4d modeling software to reconstruct a building 3D model, (3) Visualization tool for viewing images from the 3D model, (4) Window-to-wall ratio (WWR) using state-of-art deep neural network semantic segmentation, (5) Envelope thermal anomaly detection using an unsupervised machine learning clustering algorithm, and (6) Rooftop energy equipment detection based on an object detection algorithm. The testing of this method involved a comparison of additional ML-generated information overlaid on current ‘state-of-practice’ audit and remote assessment baselines using evaluation metrics: labor time and associated cost, marginal benefits of using ML-generated information in workflows for audits and remote assessments, integration potential with existing processes and tools, and replicability/scalability of the method. In two test buildings in California that had comprehensive drawings and meter data available, the ML method effectively generated a building footprint, envelope, rooftop equipment, WWR, and locations of envelope thermal anomalies. Projected target segments of the ML method are sites with minimal drawings and energy data, and underserved sectors such as multistoried housing, disadvantaged communities, and schools for which the ML method can enable identification of building asset characteristics and prioritization of envelope retrofits and decentralized energy equipment retrofits.

Singh, Reshma↗
DOE OSTI2022

Systems and methods for powder bed additive manufacturing anomaly detection

Detection and classification of anomalies for powder bed metal additive manufacturing. Anomalies, such as recoater blade impacts, binder deposition issues, spatter generation, and some porosities, are surface-visible at each layer of the building process. A multi-scaled parallel dynamic segmentation convolutional neural network architecture provides additive manufacturing machine and imaging system agnostic pixel-wise semantic segmentation of layer-wise powder bed image data. Learned knowledge is easily transferrable between different additive manufacturing machines. The anomaly detection can be conducted in real-time and provides accurate and generalizable results.

Scime, Luke R.↗
DOE OSTI2023

Compute-Efficient Deep Learning: Algorithmic Trends and Opportunities

Although deep learning has made great progress in recent years, the exploding economic and environmental costs of training neural networks are becoming unsustainable. To address this problem, there has been a great deal of research on algorithmically-efficient deep learning, which seeks to reduce training costs not at the hardware or implementation level, but through changes in the semantics of the training program. In this paper, we present a structured and comprehensive overview of the research in this field. First, we formalize the algorithmic speedup problem, then we use fundamental building blocks of algorithmically efficient training to develop a taxonomy. Our taxonomy highlights commonalities of seemingly disparate methods and reveals current research gaps. Next, we present evaluation best practices to enable comprehensive, fair, and reliable comparisons of speedup techniques. To further aid research and applications, we discuss common bottlenecks in the training pipeline (illustrated via experiments) and offer taxonomic mitigation strategies for them. Finally, we highlight some unsolved research challenges and present promising future directions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Are System Baselines within OT Environments Feasible?

Critical infrastructure stakeholders need to baseline their systems to understand expected protocol communications.Baseline behaviors may vary based on operational context.Expected operations during a maintenance window, for example, may be different from normal operations.Furthermore, constructing system baselines for Industrial Control Systems (ICS) is difficult and time-consuming.ICS processes generate artifacts expressed across heterogeneous data sources such as network and device logs. There needs to be a corpus of data in order to develop and compare methods that evaluate the feasibility, performance, and generality of approaches to construct baselines for ICS events. Standalone repositories of network packet captures are insufficient to develop methods to classify or recognize operational events expressed across multiple data sources. Moreover, static data corpora do not enable researchers to compare the impact of changing the underlying system for which a baseline is being constructed and this limits the ability to evaluate the performance of system baselines given system changes (e.g. patches, configuration, maintenance events). In order to address these limitations within the community, this talk intends to promote discussion about the state of the practice of constructing baselines. In this manner, we can continue to understand requirements within industry that are not being met by current approaches to baseline construction. This talk builds on two previous talks on the topic of system baselines for OT environments. First, Weaver co-presented at the RSA Conference ICS Sandbox with Dan Gunter. The talk confirmed the need within industry to construct baselines across multiple types of data sources relative to the semantics of specific business processes. Second, Weaver presented at IEEE Security and Privacy Workshop on Language-Theoretic Security.

02 PETROLEUM↗
DOE OSTI2012

A verification regime for the spatial discretization of the SN transport equations

The order-of-accuracy test in conjunction with the method of manufactured solutions is the current state of the art in computer code verification. In this work we investigate the application of a verification procedure including the order-of-accuracy test on a generic SN transport solver that implements the AHOTN spatial discretization. Different types of semantic errors, e.g. removal of a line of code or changing a single character, are introduced randomly into the previously verified S{sub N} code and the proposed verification procedure is used to identify the coding mistakes (if possible) and classify them. Itemized by error type we record the stage of the verification procedure where the error is detected and report the frequency with which the errors are correctly identified at various stages of the verification. Errors that remain undetected by the verification procedure are further scrutinized to determine the reason why the introduced coding mistake eluded the verification procedure. The result of this work is that the verification procedure based on an order-of-accuracy test finds almost all detectable coding mistakes but rarely, 1.44% of the time, and under certain circumstances can fail. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2013

Societal-Equity-Enhancing Criteria and Facility-Host Incentives Supporting Five Key Elements in the January 2012 Blue Ribbon Commission Report - 13015

In February 2009, the Obama Administration announced it would abandon USA's only candidate SNF/HLW-disposal facility since 1987. In 2010, all related activities were stopped and the Blue Ribbon Commission on America's Nuclear Future was established 'to recommend a new strategy for managing the back end of the nuclear fuel cycle', which it did in January 2012, emphasizing eight key elements. However, Key Element 1, 'A new, consent-based approach to siting future nuclear facilities', is qualitative/indeterminate rather than quantitative/measurable. It is thus highly-susceptible to semantic permutations that could extend rather than, as intended, expedite the siting of future nuclear facilities unless it also defines: a) Whose consent is needed?; and b) What constitutes consent? The following 'generic', radiation-risk- and societal-equity-based criteria address these questions: 1. Identify areas affected by projected radiation and other health risks from: a. The proposed nuclear facility (facility stakeholders); and b. The related nuclear-materials-transportation routes (transportation stakeholders); then 2. Surround each stakeholder area with a buffer zone and use this enlarged foot print to identify: a. Stakeholder hosts; and b. Areas not hosting any stakeholder category (interested parties). 3. Define 'consent-based' as being at least 60 percent of the 'population' in the respective stakeholder category and apply this yardstick to both 'in favor' and 'against' votes. Although criteria 1 and 2 also need facility-based definitions to make Key Element 1 measurable, the described siting approach, augmented by related facility-host incentives, would expedite the schedule and reduce the cost for achieving Key Elements 4-6 and 8, politics permitting. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

Freewheeling: What Six Locations, 61,000 Trips, and 242,000 Miles in Colorado Reveal about How E-Bikes Improve Mobility Options

Personal micromobility modes such as bicycles, e-bikes and scooters offer low- or zero-emission transportation alternatives to single occupancy vehicles (SOVs). However, the lack of supporting data has led to a dearth of data-driven research on the usage of personally owned e-bikes, including variations due to weather, geography and demographics. In this paper, we present an overview of the longitudinal findings from the CanBikeCO program, focused on e-bike adoption and use rates across different demographics, trip characteristics, and geographies. The CanBikeCO program recorded travel survey data from late July 2021 to December 31, 2022, from low-income Colorado households who were provided with e-bikes for personal use by the Colorado Energy Office (CEO). This data was collected in six different communities across Colorado following the mini-pilot program that was conducted in Fall 2020. To collect data for the survey, the program used the NREL OpenPATH application, which combines passive data collection with semantic information such as trip mode and purpose labels. To the best of our knowledge, there is no prior travel survey data on personally owned e-bikes with this range and scope. This unique dataset yielded several insights. One is that commute trips among participants had nearly 17% higher shares of e-bikes than all trips combined. E-bikes were stated to most often replace cars (34% of e-bike trips) and personal micromobility (22%). Participants favored walking for trips less than 1 mile, e-bikes for trips 1-3 miles, and e-bikes, cars or shared rides for trips 3-20 miles. Seasonality accounted for a 10% decrease and subsequent recovery in e-bike mileage on a per user basis. E-bikes are also appealing across age groups, even among older individuals, and see decreased utilization similar to regular bikes or walking during winter months. We also find that e-bike use may be related to characteristics of land use and urban form, occupation and income as well as household car ownership. We conclude that, for this population, who are mainly part of low-income households, the emissions added by the use of e-bikes (in the case of replacement of non-motorized modes) are outweighed by the strong single occupancy vehicle (SOV) travel replacement. As a whole, our findings suggest a considerable potential for energy savings and emissions reductions from personal e-bike ownership.

ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2018

Can we use Pictograms in a Long Term Communication System Related to a Radioactive Waste Repository? - 18301

Pictograms are often mentioned as a possible tool for sending messages about radioactive waste repositories in a far future, when the understand ability of our languages will have been lost. But they have also limits, related to the use of conventions that may fade over time. The goals of a signaling system for the long term may be (1) to alert (raise attention), (2) to inform of an underlying danger due to the existence of radioactive wastes, (3) to induce a favorable behavior (provoke or deter actions) in order not to disturb the performance of the repository. The quest for trans-cultural and trans-generational performance requires that some universal semiotic features be identified, shared by human beings of various cultures nowadays, and from far in the past to the present. Considering the typology of colors in various languages, the use of colors in ancient rock art and in children's drawings, a dedicated role may be assigned to each of the three colors, white, black and red/ocher, with a high potential for sustainable meaning: 1) red background = pay attention to this object because of the simple presence of the red color, that is to say the first color felt as color; 2) white inscription = absence of the object / action represented; 3) black inscription = presence of the object / action represented. The representation of a human body in movement inside a context is also understandable by people from various cultures. The combination of these 'semiotic primitives', used as relevant semantic features of the object and/or the experience to be represented, and 'topo-sensitive' criteria, used for pictogram production processes, would thus become a way to conceive a trans-cultural and trans-generational sign system for the signaling of radioactive danger. The choice of the stone could guarantee the durability of the support; the color could guarantee its visibility; and both could guarantee the transmission of the message. Combining these criteria could open the way to a long term communication system about radioactive waste repositories. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2024

MFANS 2024 - Formally Proving Characteristics of Cyber-Physical Systems

Cyber-physical systems (CPS) are engineered systems that rely on the smooth integration of computational algorithms and physical elements. This integration presents new challenges for verifying that systems will behave as expected. The goal of this presentation is to present current challenges and potential solutions for the formal verification of cyber-physical systems. For cyber systems, formal methods refer to systematically rigorous mathematical techniques employed in the specification, development, analysis, and verification of both software and hardware systems. Recent advancements in computer science have yielded sophisticated tools specifically designed to address challenges associated with formal methods in complex systems. These tools leverage various foundational concepts such as logic, formal languages, program semantics, type systems, type theory, and automata theory. A notable achievement in the application of formal methods is the seL4 microkernel, claimed to be the first general-purpose operating-system kernel to be verified. Its proof implies the absence of bugs and guarantees that the kernel meets specifications. For physical systems, dynamic and control theory has a history of using rigorous analytic techniques to prove functional correctness. Lyapunov, optimal, classical, modern, and robust control theories all provide rigorous mathematical methods both to analyze system performance and to design controller that can be guaranteed to meet certain objectives. Recent computational techniques like level set theory and reachability analysis provide assertions that a system's state will avoid unsafe regions. Even though success has been independently achieved for cyber systems and physical systems, the integration of such systems creates new challenges. In particular, there is an obvious discrepancy between finite-state machines and infinite-state systems, resulting in different approaches for modeling and analyzing these system. While it is possible to simulate hybrid systems, this provides only a demonstration of a performance and not proof. For hybrid systems, current formal methods and system analysis approaches typically require a workarounds to work on hybrid systems like CPS. This paper will outline the state of the art and limits of current practice for formally verifying CPS and will identify possible research directions that require attention.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

GLIMPSE of Future Power Grid Models

Power grid is one of the critical national infrastructures with social, economics, and national security impacts. Particularly, power distribution systems represent part of the infrastructure between power distribution substations and customers such as residential, commercial, and industrial. To address various grid modernization challenges, state-of-the-art algorithms and methodologies have been developed to deploy, operate, and expand a secure and resilient power grid. At the same time, there is a need to develop capabilities to assist power grid stakeholders to quickly get insight into the design and structure of the grid. Data visualization is a key approach to comprehend and understand complex systems such as the power grid. We present GLIMPSE Grid Layout Interface for Model Preview and System Exploration), a graph-based semantic-aware application to visualize and update distribution power grid models. The GLIMPSE can be used with standard IEEE models to search and highlight power grid objects such as generators, loads, overhead lines, etc. Additionally, it supports updating attributes and model export to integrate with GridLAB-D simulations.

Cybersecurity, power grid, visualization↗
DOE OSTI2024

Uncertainty-aware Continuous Implicit Neural Representations for Remote Sensing Object Counting

Many existing object counting methods rely on density map estimation (DME) of the discrete grid representation by decoding extracted image semantic features from designed convolutional neural networks (CNNs). Relying on discrete density maps not only leads to information loss dependent on the original image resolution, but also has a scalability issue when analyzing high-resolution images with cubically increasing memory complexity. Furthermore, none of the existing methods can offer reliable uncertainty quantification (UQ) for the derived count estimates. To overcome these limitations, we design UNcertainty-aware, hypernetwork-based Implicit neural representations for Counting (UNIC) to assign probabilities and the corresponding counting confidence over continuous spatial coordinates. We derive a sampling-based Bayesian counting loss function and develop the corresponding model training algorithm. UNIC outperforms existing methods on the Remote Sensing Object Counting (RSOC) dataset with reliable UQ and improved interpretability of the derived count estimates. Our code is available at https://github.com/SiyuanXu-tamu/UNIC.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Systems and methods for binary code analysis

Human-readable (HR) code may be derived from a binary. The HR code may be configured to have statistical properties suitable for machine-learned (ML) translation. The HR code may comprise source code, intermediate code, assembly code, or the like. A machine-learned translator may be configured to translate the HR code into labels comprising semantic information pertaining to respective functions of the binary, such as a function name, role, or the like. Execution of the binary may be blocked in response to translating the HR code to a label associated with malware, such as cryptocurrency mining malware or the like. Conversely, the binary may be permitted to proceed to execution in response to determining that the translation is free from labels indicative of malware.

Anderson, Matthew W.↗
DOE OSTI2025

MoRE-Brain: Routed Mixture of Experts for Interpretable and Generalizable Cross-Subject fMRI Visual Decoding

Decoding visual experiences from fMRI offers a powerful avenue to understand human perception and develop advanced brain-computer interfaces. However, current progress often prioritizes maximizing reconstruction fidelity while overlooking interpretability, an essential aspect for deriving neuroscientific insight. To address this gap, we propose MoRE-Brain, a neuro-inspired framework designed for high-fidelity, adaptable, and interpretable visual reconstruction. MoRE-Brain uniquely employs a hierarchical Mixture-of-Experts architecture where distinct experts process fMRI signals from functionally related voxel groups, mimicking specialized brain networks. The experts are first trained to encode fMRI into the frozen CLIP space. A finetuned diffusion model then synthesizes images, guided by expert outputs through a novel dual-stage routing mechanism that dynamically weighs expert contributions across the diffusion process. MoRE-Brain offers three main advancements: First, it introduces a novel Mixture-of-Experts architecture grounded in brain network principles for neuro-decoding. Second, it achieves efficient cross-subject generalization by sharing core expert networks while adapting only subject-specific routers. Third, it provides enhanced mechanistic insight, as the explicit routing reveals precisely how different modeled brain regions shape the semantic and spatial attributes of the reconstructed image. Extensive experiments validate MoRE-Brain’s high reconstruction fidelity, with bottleneck analyses further demonstrating its effective utilization of fMRI signals, distinguishing genuine neural decoding from over-reliance on generative priors. Consequently, MoRE-Brain marks a substantial advance towards more generalizable and interpretable fMRI-based visual decoding.

Wei, Yuxiang [Georgia Institute of Technology]↗
DOE OSTI2025

Can Large Language Models Understand Intermediate Representations?

Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.

Jiang, Hailong↗
DOE OSTI2026

Agentic Diagrammatica: Towards Autonomous Symbolic Computation in High Energy Physics

We present Diagrammatica, a symbolic computation extension to the HEPTAPOD agentic framework, which enables LLM agents to plan and execute multi-step theoretical calculations. Symbolic computation poses a distinctive reliability challenge for LLM agents, as correctness is governed by implicit mathematical conventions that are not encoded in a form that can be easily checked in the computational backend. We identify two complementary remedies, tool-constrained computation and targeted knowledge grounding, and pursue the first as the primary architecture. Concretely, we concentrate the agent's action distribution onto tool calls with convention-fixing semantics, in which the agent specifies a compact, human-auditable diagram specification and a trusted backend performs the symbolic or numerical manipulations exactly. The toolkit provides two complementary calculation paths consuming a shared diagram specification: Naive Dimensional Analysis (NDA) for order-of-magnitude rate estimates and Exact Diagrammatic Analysis (EDA) for tree-level symbolic calculations via automatic FeynCalc code generation, both supplemented by automatic Feynman diagram enumeration and a navigable theory knowledge base. The architecture is validated on two benchmarks: (1) an exhaustive catalog of all tree-level, single-vertex $1\to 2$ partial decay widths across scalar, fermion, and vector parents, with complete massless and threshold limits and Standard Model validation; and (2) an NDA sensitivity study of the muon decay multiplicity $μ^+ \to ν_μ\barν_e + n(e^+e^-) + e^-$, determining the maximum observable $n$ at current and planned muon experiments.

Menzo, Tony [Alabama U.; Fermilab] (ORCID:00000002↗
DOE OSTI2026

Resource-Adaptive Federated Text Generation with Differential Privacy

In cross-silo federated learning (FL), sensitive text datasets remain confined to local organizations due to privacy regulations, making repeated training for each downstream task both communication-intensive and privacy-demanding. A promising alternative is to generate differentially private (DP) synthetic datasets that approximate the global distribution and can be reused across tasks. However, pretrained large language models (LLMs) often fail under domain shift, and federated finetuning is hindered by computational heterogeneity: only resource-rich clients can update the model, while weaker clients are excluded, amplifying data skew and the adverse effects of DP noise. We propose a flexible participation framework that adapts to client capacities. Strong clients perform DP federated finetuning, while weak clients contribute through a lightweight DP voting mechanism that refines synthetic text. To ensure the synthetic data mirrors the global dataset, we apply control codes (e.g., labels, topics, metadata) that represent each client’s data proportions and constrain voting to semantically coherent subsets. This two-phase approach requires only a single round of communication for weak clients and integrates contributions from all participants. Experiments show that our framework improves distribution alignment and downstream robustness under DP and heterogeneity.

Wang, Jiayi [ORNL]↗
DOE OSTI2026

Selective Amnesia using Contrastive Subnet Erasure for Class Level Unlearning in Vision Models

We study concept-level forgetting in pretrained vision models: removing an entire semantic category so the system no longer recognizes that object in unseen images and contexts, rather than merely forgetting specific training examples. Prior work either applies blunt global projections or fine-tunes parameters, which can introduce collateral damage to unrelated features, add compute, and become unstable as forgetting strength increases. We introduce Contrastive Subnet Erasure (CSE), a training-free, encoder-centric edit that targets a compact set of channels most responsible for the class and attenuates them in a calibrated manner. The modification is algebraically folded into the subsequent layer, yielding no inference-time overhead and leaving task heads unchanged. To evaluate whether forgetting generalizes beyond the data used to specify the class, we introduce a cross dataset protocol in which the class is defined on a source dataset and performance is measured on a disjoint target dataset drawn from a different distribution with no shared images. This setup tests whether the model still fails to recognize the object when it looks different or appears in new scenes, and it helps avoid overfitting to patterns in the source dataset. Across CIFAR 10, CIFAR 100, and ImageNet under this protocol, CSE achieves stronger forgetting of the target class while better preserving non target utility than existing baselines in both single class and multi class settings. Overall, CSE provides a simple, stable, and deployment-ready mechanism for class-level unlearning in vision.

Kotevska, Olivera [ORNL] (ORCID:0000000316772243)↗