Search NASA⌕ Search

SEARCH · Search NASA

Results for “graph learning”

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 865 records · Page 48

Training courses on neutron detection systems on the ISIS research reactor: on-site and through internet training

Today, ISIS research reactor is an essential tool for Education and Training programs organized by the National Institute for Nuclear Science and Technology (INSTN) from CEA. In the field of nuclear instrumentation, the INSTN offers both, theoretical courses and training courses on the use of neutron detection systems taking advantage of the ISIS research reactor for the supply of a wide range of neutron fluxes. This paper describes the content of the training carried out on the use of neutron detectors and detection systems, on-site or remote. The ISIS reactor is a 700 kW open core pool type reactor. The facility is very flexible since neutron detectors can be inserted into the core or its vicinity, and be used at different levels of power according to the needs of the course. Neutron fluxes, typically ranging from 1 to 10{sup 12} n/cm{sup 2}.s, can be obtained for the characterisation of the neutron detectors and detection systems. For the monitoring of the neutron density at low level of power, the Instrumentation and Control (I and C) system of the reactor is equipped with two detection systems, named BN1 and BN2. Each way contains a fission chamber, type CFUL01, connected to an electronic system type SIREX.The system works in pulse mode and exhibits two outputs: the counting rate and the doubling time. For the high level of power, the I and C is equipped with two detection systems HN1 and HN2.Each way contain a boron ionization chamber (type CC52) connected to an electronics system type SIREX. The system works in current mode and has two outputs: the current and the doubling time. For each mode, the trainees can observe and measure the signal at the different stages of the electronic system, with an oscilloscope. They can understand the role of each component of the detection system: detector, cable and each electronic block. The limitation of the detection modes and their operating range can be established from the measured signal. The trainees can also modify the settings of the electronic system, such as the high voltage and the discrimination level in order to obtain all the characteristic curves of the detectors. These curves are used to define the right setting of the electronic system and to discuss the expected degradation of the detector signal resulting from the detector damage under the integrated neutron and gamma fluxes. Moreover, in addition to the study of the neutron detection systems itself, the integration of the measurements made by these detection systems in the logic of the safety system of the nuclear reactor is also addressed. Providing the trainees with an extensive overview of each part of the neutron monitoring instrumentation apply to a nuclear reactor, hands-on measurements on the ISIS reactor play a major role in ensuring a practical and comprehensive understanding of the neutron detection system and their integration in the safety system of nuclear reactors. It also gives a solid background for the follow up and the development of the neutron detection systems. In addition to on-reactor training, Internet Reactor Laboratory capability has been implemented on the ISIS reactor in 2014. For the Internet Reactor Laboratory an extensive video conference system has been implemented on ISIS reactor. The system includes 4 cameras and the transmission of the video signal given by the supervision system of the reactor which records and processes the data of the reactor. According to the pedagogic needs during the training courses, the lecturer on the ISIS reactor chooses to broadcast the relevant information at each stage of the course. For example, graph showing the histogram of the counting and current as a function of the time, or the electrical signal observed on the oscilloscope, can be broadcasted trough internet. By interacting through the video conference, the remote classroom is able to ask for changes in the reactor power or settings of the detection systems. They can also ask for the broadcast of some particular information. At the guest institution, the information is displayed in two parts or screens, as shown in the Figure 3. Concerning the interaction with - and the feedback from - the remote classroom, the camera of the video system in the remote classroom is used to ensure the contact between the trainees and the lecturer and reactor operators. Thus, the Internet Reactor Laboratory is complementary to the on reactor training courses. It allows distant learning, reducing the overall cost of the course when this is necessary. It can efficiently be used for the development of the human resources needed by the nuclear industry and the nuclear programs in countries without research reactors.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Machine Learning Approaches for Nuclear Material Accounting Data from Irradiation and Reprocessing

We are currently exploring data analysis methods for their ability to strengthen the synthesis and evaluation of information generated within domestic and international safeguard regimes. Safeguard data typically includes rich heterogenous datasets amenable to advanced data analytics methods, such as machine learning. We have converted transactional data containing both numerical and categorical attributes from a domestic nuclear material control and accountability (NMC&A) system into a low-dimensional numerical structure via linear principal component analysis. This data representation allows for global structure discovery via cluster analysis, which can characterize the typical behavior of each of the primary types of transaction events. Furthermore, the structure of the top principal components captures the “typical behavior” of the data and is thus amenable to anomaly detection through statistical hypothesis testing. We explored this capability by generating erroneous permutations of the data and computing the Q-residual quantity increase associated with the information loss when these data are projected into the low-dimensional principal component analysis space representative of typical transactions.Future work will focus on identifying data transformations (e.g., graph networks) that more closely align with the inherent structure of these transactions to explain more salient information and disambiguate the underlying nuclear process—in this case, irradiation and reprocessing—from artifacts of NMC&A system transactional record keeping data.

Drescher, Adam↗

InteraChem: Virtual Reality Visualizer for Reactive Interactive Molecular Dynamics

Interactive molecular dynamics in virtual reality (IMD-VR) simulations provide a digital molecular playground for students as an alternative or complement to traditional molecular modeling kits or 2D illustrations. Previous IMD-VR studies have used molecular mechanics to enable simulations of macromolecules such as proteins and nanostructures for the classroom setting with considerable success. Here, we present the InteraChem molecular visualizer, intended for reactive IMD-VR simulation using semiempirical and ab initio methods. InteraChem visualizes not only the molecular geometry but also (1) isosurfaces such as molecular orbitals and electrostatic potentials and (2) two-dimensional graphs of time-varying simulation quantities such as kinetic/potential energy, internal coordinates, and user-applied force. Additionally, InteraChem employs speech recognition to facilitate user interaction and introduces a novel “atom happiness” visualization using emojis to indicate the energetic feasibility of a particular bonding arrangement. Furthermore, we include a set of accompanying exercises that we have used to teach chemical reactivity in small molecular systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)

The development of modern processors exhibits two trends that complicate the optimizations of modern software. The first is the increasing sensitivity of processors' throughput to irregularities in computation. With more processors produced through a massive integration of simple cores, future systems will increasingly favor regular data-level parallel computations, but deviate from the needs of applications with complex patterns. Some evidences are already shown on Graphic Processing Units (GPU): Irregular data accesses (e.g., indirect references A[D[i]]) and conditional branches are limiting many GPU applications' performance at a level an order of magnitude lower than the peak of GPU. The second hardware trend is the growing gap between memory bandwidth and the aggregate speed—that is, the sum of all cores' computing power—of a Chip Multiprocessor (CMP). Despite the capped growth of the peak CPU speed, the aggregate speed of a CMP keeps increasing as more cores get into a single chip. It is expected that by 2018, node concurrency in an exascale system will increase by hundreds of times, whereas, memory bandwidth will expand by only 10 to 20 times. Consequently, data movement and storage is expected to consume more than 70% of the total system power. Bridging this gap is difficult; the complexities of modern CMP memory hierarchy make it even harder: Data cache becomes shared among computing units, and the sharing is often non-uniform—whether two computing units share a cache depends on their proximity and the level of the cache. On the recent IBM Power7 architecture, for instance, four hardware contexts (or SMT threads) in a core share the entire memory hierarchy, all cores in one chip share an on-chip L3 cache, and cores across chips share L3 and main memory through off-chip connections. These two trends complicate the translation of computing power into performance, especially for a program with either intensive data accesses or complex patterns in data accesses or control flow paths. Unfortunately, both attributes present and will persist in a class of important applications. For instance, many scientific simulations deal with a large volume of data. And meanwhile, as most real-world processes are non-uniform and evolving (e.g., the evolution of a galaxy or the process of a drug injection), both the computations and data accesses of these programs tend to be irregular and dynamically changing. Currently, the lack of support to these applications on modern CMP severely limits their performance. On GPU, as our recent study shows and other studies echo, performance enhancement of a factor of integers is possible when memory accesses or control flows are streamlined for a set of GPU applications. On multicore CPU, our studies show that traditional locality enhancement, for being oblivious to the new features of multicore memory hierarchy, may even cause large slowdown to data-intensive dynamic applications. The severity of the issues is expected to worsen as the two hardware trends continue. Some recent studies try to match software with the trends, but in a limited scope or manner. For irregularities on GPU, most studies focus on irregularities analyzable through static analysis (e.g., data accesses in regular loops). Dynamic irregularities are harder to address because the needed analysis and transformations typically have to happen at run time. Some other research resorts to hardware extensions, an actual adoption of which is unclear for the entailed space cost and complexity. For data locality, recent years have seen some exploitations of the new memory hierarchy on multicore for performance, but most of them are on process or thread scheduling, rather than program transformations. Our recent study reveals that program-level transformations may magnify the scheduling benefits by a factor of seven, concluding that program-level transformation should play a central role for data locality enhancement on modern CMP. But research in this direction has been sparse, and most have focused on data layout or cache performance modeling, rather than program transformations to match with the new memory hierarchy features. Overall, it is still an open question how to bridge the gap between dynamic computations and the two prominent properties of modern processors. The goal of this project is to develop a set of techniques and software tools to enhance the matching between memory accesses in dynamic simulations and the prominent features of modern and future CMP, alleviating the memory performance issues for petascale and exascale computing. This report summarizes the discoveries and products produced throughout this project. It includes free launch, a new software approach to overcoming the shortcomings of both methods; coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution; algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Supporting dynamic parallelism is important for GPU to benefit a broad range of applications. There are currently two fundamental ways for programs to exploit dynamic parallelism on GPU: a software-based approach with software-managed worklists, and a hardware-based approach through dynamic subkernel launches. Neither is satisfactory. The former is complicated to program and is often subject to some load imbalance; the latter suffers large runtime overhead. In this work, we propose free launch, a new software approach to overcoming the shortcomings of both methods. It allows programmers to use subkernel launches to express dynamic parallelism. It employs a novel compiler-based code transformation named subkernel launch removal to replace the subkernel launches with the reuse of parent threads. Coupled with an adaptive task assignment mechanism, the transformation reassigns the tasks in the subkernels to the parent threads with a good load balance. The technique requires no hardware extensions, immediately deployable on existing GPUs. It keeps the programming convenience of the subkernel launch-based approach while avoiding its large runtime overhead. Meanwhile, its superior load balancing makes it outperform manual worklist-based techniques by 3X on average. The work was published at the 48th Annual IEEE/ACM International Symposium on Microarchitecture (Micro'2015). A Graphic Processing Unit (GPU) system is typically equipped with many types of memory (e.g., global, constant, texture, shared, cache). Data placement determines what data are placed on which type of memory, essential for GPU memory performance. Prior optimizations of data placement always require a single view of a data object on memory, which limits the optimization effectiveness. In this work, we propose coherence-free multiview, an approach that allows multiple views of a single data object to co-exist on GPU memory during a GPU kernel execution. We demonstrate that under certain conditions, the multiple views can remain incoherent while facilitating enhanced data placement. We present a theorem and some compiler support to ensure the soundness of the usage of coherence-free multiview. We further develop reference-discerning data placement, a new way to enhance data placements on GPU. It enables more flexible data placements by using coherence-free multiview to leverage the slack in coherence requirement of some GPU programs. Experiments on three types of GPU systems show that, with less than 200KB space cost, the new data placement technique can provide a 1.6X average (up to 4.27X) speedup. In addition, we have examined algorithmic optimizations to data analytics problems, especially those that involve lots of distance calculations. Computing distances among data points is an essential part of many important algorithms in data analytics, graph analysis, and other domains. In each of these domains, developers have spent significant manual effort optimizing algorithms, often through novel applications of the triangle equality, in order to minimize the number of distance computations in the algorithms. In this work, we observe that many algorithms across these domains can be generalized as an instance of a generic distance-related abstraction. Based on this abstraction, we derive seven principles for correctly applying the triangular inequality to optimize distance-related algorithms. Guided by the findings, we develop Triangular OPtimizer (TOP), the first software framework that is able to automatically produce optimized algorithms that either matches or outperforms manually designed algorithms for solving distance-related problems. TOP achieves up to 237x speedups and 2.5X on average. The work has been published at the 32nd International Conference on Machine Learning (ICML'15) and the 41st International Conference on Very Large Data Bases (VLDB'15).

42 ENGINEERING↗

Potential Use of a Bayesian Network for Discriminating Flash Type from Future GOES-R Geostationary Lightning Mapper (GLM) data

Continuous monitoring of the ratio of cloud flashes to ground flashes may provide a better understanding of thunderstorm dynamics, intensification, and evolution, and it may be useful in severe weather warning. The National Lighting Detection Network TM (NLDN) senses ground flashes with exceptional detection efficiency and accuracy over most of the continental United States. A proposed Geostationary Lightning Mapper (GLM) aboard the Geostationary Operational Environmental Satellite (GOES-R) will look at the western hemisphere, and among the lightning data products to be made available will be the fundamental optical flash parameters for both cloud and ground flashes: radiance, area, duration, number of optical groups, and number of optical events. Previous studies have demonstrated that the optical flash parameter statistics of ground and cloud lightning, which are observable from space, are significantly different. This study investigates a Bayesian network methodology for discriminating lightning flash type (ground or cloud) using the lightning optical data and ancillary GOES-R data. A Directed Acyclic Graph (DAG) is set up with lightning as a "root" and data observed by GLM as the "leaves." This allows for a direct calculation of the joint probability distribution function for the lighting type and radiance, area, etc. Initially, the conditional probabilities that will be required can be estimated from the Lightning Imaging Sensor (LIS) and the Optical Transient Detector (OTD) together with NLDN data. Directly manipulating the joint distribution will yield the conditional probability that a lightning flash is a ground flash given the evidence, which consists of the observed lightning optical data [and possibly cloud data retrieved from the GOES-R Advanced Baseline Imager (ABI) in a more mature Bayesian network configuration]. Later, actual GLM and NLDN data can be used to refine the estimates of the conditional probabilities used in the model; i.e., the Bayesian network is a learning network. Methods for efficient calculation of the conditional probabilities (e.g., an algorithm using junction trees), finding data conflicts, goodness of fit, and dealing with missing data will also be addressed.

Solakiewiz, Richard↗

The Process of the Development of an Operator

On the job training is where new employees called operators can start gaining knowledge on what they will be working on during their time in JSC. In these lessons I learned different things that are important ranging from thermal systems to electrical systems. While doing OJT classes the student will learn how to use a portable computer system which has displays that I also helped edit and clean up. The way you can also learn is by reading system briefs which describes the different systems. Due to the fact of a possible change in the ISS I updated a systems brief so that it can be relevant to what is actually on the space station. I was given a task that will help develop my skills and make myself better prepared for my future in the work field. The project that I worked on had me pulling real time data from the International Space Station. The Data I obtained from the space station will be correlated to battery performance. The group I will be working which is called REBA and we will take the telemetry and evaluate the data. I will be working with my mentor Ben Chislom and co-op Tyler along with the Pro team. They then will put this data into a graph so that they can get the discrepancies and find a way to improve the battery performance. The first weeks I read familiarization books that informed me how the ISS works, how it was built, and the systems that are used to keep the station working. This project is going to benefit NASA by finding out how electricity is being used on the ISS and enabling us to see how it can be used more efficiently. This way we can operate the ISS without wasting power. While conducting research that goes on inside the space station knowing all electricity is being used efficiently.

Banks, Terrence↗

LLMs for Mfg.—On the State of Large Language Models and Applications to Manufacturing

Additive Manufacturing (AM), referred to as 3D printing, has emerged as a key pillar of Industry 4.0 enabling layer-by-layer fabrication of intricate geometries from CAD models. In parallel, Large Language Models (LLMs), deep learning models for natural language generation trained on vast text corpora, have demonstrated unprecedented capabilities in understanding and generating human-like text. The convergence of these trends opens new opportunities at the intersection of AM and AI/ML, where LLMs can assist engineers and researchers in design, manufacture planning, and knowledge discovery. Recent academic work has begun to explore LLM applications in AM and adjacent fields, such as material science, mechanical engineering, and design for additive manufacturing. This exploration ranges from intelligent process planning to domain-specific knowledge retrieval. This survey provides a comprehensive review of current developments, focusing on peer-reviewed literature contributions that apply, adapt, and advance LLMs in general and domain-specific domains. We analyze state-of-the-art (SOTA) techniques, such as fine-tuning foundational models for specific domains, retrieval-augmented generation (RAG) pipelines, knowledge graph integration, and delve into the architectures and evaluation methods employed. The goal of this survey is to inform researchers and practitioners of the current capabilities and limitations of LLMs in general and in domain-specific applications, and to outline how these models are being tailored to meet the requirements of these applications.

36 MATERIALS SCIENCE↗

PySIDT: Subgraph Isomorphic Decision Trees for Molecular Property Prediction

Accurate molecular property prediction is important across all fields of chemistry. Deep neural networks (DNNs) have become increasingly popular due to their ability to train automatically, avoiding the incredibly tedious process of constructing and extending traditional property estimation schemes. However, DNNs require large amounts of training data, are challenging to interpret, require large amounts of memory to load even during inference, and have severe difficulties incorporating qualitative chemical knowledge, which are often desired for molecular property prediction tasks. Here, in this study, we present PySIDT (https://github.com/zadorlab/PySIDT), a software for training and running inference on Subgraph Isomorphic Decision Trees (SIDTs). SIDTs are graph-based decision trees made of nodes associated with molecular substructures. Inference is done by descending target molecular structures down the decision tree to nodes with matching subgraph isomorphic substructures and making predictions based on the final (most specific) nodes matched. SIDTs scale down well to dataset sizes much smaller than is feasible for DNNs. As trees of molecular substructures, SIDTs are inherently readable and easy to visualize, making them easy to analyze. They are also straightforward to extend and retrain, facilitate uncertainty estimation, and enable easy integration of expert knowledge. We demonstrate the SIDT approach discussing its application to a diverse range of molecular prediction tasks: rate coefficient estimation, diffusion coefficient estimation, thermochemistry estimation, transition state bond stretch prediction, p K a prediction, stability of molecular structures, stability of surface structures, and prediction of surface lateral interaction energetics. Additionally, we demonstrate the power of the SIDT algorithms in two direct learning curve vanilla comparisons with the popular DNN-based software Chemprop and the popular gradient boosted trees-based software XGBoost on enthalpy of formation and rate coefficient prediction tasks. In particular, in the enthalpy of formation case, vanilla PySIDT is able to outperform vanilla Chemprop and XGBoost across the full range of training/validation set sizes out to 11,560 data points.

Johnson, Matthew Sean [Sandia National Laboratorie↗

RAP: Resource-aware Automated GPU Sharing for Multi-GPU Recommendation Model Training and Input Preprocessing

Ensuring high-quality recommendations for newly onboarded users requires the continuous retraining of Deep Learning Recommendation Models (DLRMs) with freshly generated data. To serve the online DLRM retraining, existing solutions use hundreds of CPU computing nodes designated for input preprocessing, causing significant power consumption that surpasses even the power usage of GPU trainers. To this end, we propose RAP, an end-to-end DLRM training framework that supports Resource-aware Automated GPU sharing for DLRM input Preprocessing and Training. The core idea of RAP is to accurately capture the remaining GPU computing resources during DLRM training for input preprocessing, achieving superior training efficiency without requiring additional resources. Specifically, RAP utilizes a co-running cost model to efficiently assess the costs of various input preprocessing operations, and it implements a resource-aware horizontal fusion technique that adaptively merges smaller kernels according to GPU availability, circumventing any interference with DLRM training. In addition, RAP leverages a heuristic searching algorithm that jointly optimizes both the input preprocessing graph mapping and the co-running schedule to maximize the end-to-end DLRM training throughput. The comprehensive evaluation shows that RAP achieves 78.3× speedup on average over CPU-based DLRM input preprocessing frameworks. In addition, the end-to-end training throughput of RAP is only 2.04% lower than the ideal case, which has no input preprocessing overhead.

Wang, Zheng↗

PylotDB - A Database Management, Graphing, and Analysis Tool Written in Python

PylotDB, written completely in Python, provides a user interface (UI) with which to interact with, analyze, graph data from, and manage open source databases such as MySQL. The UI mitigates the user having to know in-depth knowledge of the database application programming interface (API). PylotDB allows the user to generate various kinds of plots from user-selected data; generate statistical information on text as well as numerical fields; backup and restore databases; compare database tables across different databases as well as across different servers; extract information from any field to create new fields; generate, edit, and delete databases, tables, and fields; generate or read into a table CSV data; and similar operations. Since much of the database information is brought under control of the Python computer language, PylotDB is not intended for huge databases for which MySQL and Oracle, for example, are better suited. PylotDB is better suited for smaller databases that might be typically needed in a small research group situation. PylotDB can also be used as a learning tool for database applications in general.

Barnette, DanielW.↗

Incorporation of Human Risk Directed Acyclic Graphs (DAG) With Mishap Investigations to Un-Silo Knowledge

NASA’s Human System Risk Board (HSRB) has been a central driver in efforts to understand, mitigate, and communicate the 29 human systems risks monitored by the board. As a result of the collaboration between research, operations, and technical authorities, large bodies of knowledge have been collected and digested to represent the current understanding of the risks. As a part of these bodies of knowledge, directed acyclic graphs (DAGs) have been developed to communicate the current understanding of the causal relationship of the hazards, contributing factors, countermeasures, other risks, and outcomes that contribute to the overall risk. This risk knowledge is applied in a theoretical sense for potential incidents during exploration even while informed by surveillance data. However, there have been mishaps and close calls during past space exploration that intersect with one or more of the Human System Risks DAGs and knowledge bases. The purpose of this exercise was to un-silo this risk knowledge and connect it to the close call of EVA 23 through the development of a DAG representing the intersection of the HSRB Risks and the events of the close call. The development of the DAG occurred through an iterative process, with each iteration expanding and/or refining the nodes and connections described by the source materials. In addition to the risk documentation developed by the HSRB, lessons learned and other mishap investigation documents were utilized to understand the events that led to water entering the helmet of a crewmember on the EVA. New nodes specific to the events of EVA 23 were interconnected with existing HSRB DAG nodes and edges. Nodes within the DAG were defined within a “DAG-tionary” with any updates to a definition that may have previously existed from the HSRB DAGs, and edges were recorded in a matrix. Both the DAG-tionary and matrix describe where nodes and edges are present across the Risk and Mishap DAG. This DAG will then be reviewed by experts outside of HSRB and HRP to confirm that interpretations of the non-health related events (such as the engineering nodes) are represented accurately. DISCUSSION This process highlighted a method by which the knowledge generated among the contributing members of the HSRB Risks can be effectively adapted and utilized through the tools employed by the Risk Custodian teams. By leveraging these tools, new context and insights to the information at hand can be brought forward to address current spaceflight challenges. Moreover, un-siloing this knowledge through future DAGs and other efforts can drive interprofessional collaboration and foster communication. This will enable teams to work together more effectively, leveraging their diverse expertise to tackle the complex challenges of space exploration and human research. Ultimately, this collaboration will bring NASA closer to achieve agency goals and contribute to the overall shared mission and vision.

Samuel Jacobs↗

Codiscovering graphical structure and functional relationships within data: A Gaussian Process framework for connecting the dots

Most problems within and beyond the scientific domain can be framed into one of the following three levels of complexity of function approximation. Type 1: Approximate an unknown function given input/output data. Type 2: Consider a collection of variables and functions, some of which are unknown, indexed by the nodes and hyperedges of a hypergraph (a generalized graph where edges can connect more than two vertices). Given partial observations of the variables of the hypergraph (satisfying the functional dependencies imposed by its structure), approximate all the unobserved variables and unknown functions. Type 3: Expanding on Type 2, if the hypergraph structure itself is unknown, use partial observations of the variables of the hypergraph to discover its structure and approximate its unknown functions. These hypergraphs offer a natural platform for organizing, communicating, and processing computational knowledge. While most scientific problems can be framed as the data-driven discovery of unknown functions in a computational hypergraph whose structure is known (Type 2), many require the data-driven discovery of the structure (connectivity) of the hypergraph itself (Type 3). We introduce an interpretable Gaussian Process (GP) framework for such (Type 3) problems that does not require randomization of the data, access to or control over its sampling, or sparsity of the unknown functions in a known or learned basis. Its polynomial complexity, which contrasts sharply with the super-exponential complexity of causal inference methods, is enabled by the nonlinear ANOVA capabilities of GPs used as a sensing mechanism.

Science & Technology - Other Topics↗

Natural Language Processing-Enhanced Nuclear Industry Operating Experience Data Analysis: Aggregation and Interpretation of Multi-Report Analysis Results

Industry-wide operating experience is a critical source of raw data for reliability and risk model parameter estimations for nuclear power plants. A large portion of operating experience data are failure events stored as reports that contain unstructured data, such as narratives. In current practice, a failure report is usually reviewed and manually coded by analysts. The coding is based on extracting several event characteristics such as system name, component type, sub-part type, failure mode, and failure cause. Event narratives are mostly used to help understand events and extract their characteristics. In this line of research, we aim to maximize the usage of event narratives by leveraging natural language processing (NLP) methods to automatically convert an event narrative to a causal graph. This research has promise to improve physical understanding of failure initiation and propagation and to facilitate use of non-failure data (e.g., near-misses and degradations) to complement the limited data pool of failures. In our previous work, we developed an NLP tool and applied it to analyze a number of licensee event reports submitted by U.S. nuclear power plants to the Nuclear Regulatory Commission. In this paper, we will report our recent research progress in aggregating the results of multiple reports, developing network model(s), and drawing statistical insights.

99 GENERAL AND MISCELLANEOUS↗

From Data to Discovery: AI's Transformative Role in Thin Film Research

The advancement of thin film technologies is pivotal for progress in numerous fields, including energy, electronics, and quantum computing. However, the traditional trial-and-error approach to materials discovery is inherently slow and inefficient. This presentation will showcase how artificial intelligence (AI) is transforming thin film research by enabling a data-driven paradigm shift. We will highlight our past successes in applying AI to understand radiation damage in thin film oxides, demonstrating how graph analytics can unravel complex material behavior. Additionally, we will provide insights into our current work at the National Renewable Energy Laboratory, where we are leading the charge in autonomous materials science. Backed by a $14M investment in our characterization facility, we are developing AI-guided workflows that seamlessly integrate experimentation and AI-guided decision-making. By harnessing the power of AI, we aim to accelerate the discovery and design of high-performance thin films, propelling innovation across a multitude of industries.

36 MATERIALS SCIENCE↗

Fast and Accurate Predictions of Total Energy for Solid Solution Alloys with Graph Convolutional Neural Networks

We use graph convolutional neural networks (GCNNs) to produce fast and accurate predictions of the total energy of solid solution binary alloys. GCNNs allow us to abstract the lattice structure of a solid material as a graph, whereby atoms are modeled as nodes and metallic bonds as edges. This representation naturally incorporates information about the structure of the material, thereby eliminating the need for computationally expensive data pre-processing which would be required with standard neural network (NN) approaches. We train GCNNs on ab-initio density functional theory (DFT) for copper-gold (CuAu) and iron-platinum (FePt) data that has been generated by running the LSMS-3 code, which implements a locally self-consistent multiple scattering method, on OLCF supercomputers Titan and Summit. GCNN outperforms the ab-initio DFT simulation by orders of magnitude in terms of computational time to produce the estimate of the total energy for a given atomic configuration of the lattice structure. We compare the predictive performance of GCNN models against a standard NN such as dense feedforward multi-layer perceptron (MLP) by using the root-mean-squared errors to quantify the predictive quality of the deep learning (DL) models. We find that the attainable accuracy of GCNNs is at least an order of magnitude better than that of the MLP.

Lupo Pasini, Massimiliano↗

Exploring and Visualizing A-Train Instrument Data

The succession of US and international satellites that follow each other in close succession, known as the A-Train, affords an opportunity to atmospheric researchers that no single platform could provide: Increasing the number of observations at any given geographic location.. . a more complete "virtual science platform". However, vertically and horizontally, co-registering and regridding datasets from independently developed missions, Aqua, Calipso, Cloudsat, Parasol, and Aura, so that they can be inter-compared can be daunting to some, and may be repeated by many. Scientists will individually spend much of their time and resources acquiring A-Train datasets of interest residing at various locations, developing algorithms to match up and graph datasets along the A-Train track, and search through large amounts of data for areas and/or phenomena of interest. The aggregate amount of effort that can be expended on repeating pre-science tasks could climb into the tens of millions of dollars. The goal of the A-Train Data Depot (ATDD) is to enable free movement of remotely located A-Train data so that they are combined to create a consolidated vertical view of the Earth's Atmosphere along the A-Train tracks. The innovative approach of analyzing and visualizing atmospheric profiles along the platforms track (i.e., time) is accomplished by through the ATDDs Giovanni data analysis and visualization tool. Giovanni brings together data from Aqua (MODIS, AIRS, AMSR-E), Cloudsat (cloud profiling radar) and Calipso (CALIOP, IIR), as well as the Aura (OMI, MLS, HIRDLS, TES) to create a consolidated vertical view of the Earth's Atmosphere along the A-Train tracks. This easy to learn and use exploration tool will allow users to create vertical profiles of any desired A-Train dataset, for any given time of choice. This presentation shows the power of Giovanni by describing and illustrating how this tool facilitates and aids A-Train science and research. A web based display system Giovanni provides users with the capability of creating co-located profile images of temperature and humidity data from the MODIS, MLS and AIRS instruments for a user specified time and spatial area. In addition, Cloud and Aerosol profiles may also be displayed for the Cloudsat and Caliop instruments. The ability to modify horizontal and vertical axis range, data range and dynamic color range is also provided. Two dimensional strip plots of MODIS, AIRS, OM1 and POLDER parameters, co-located along the Cloudsat reference track, can also be plotted along with the Cloudsat cloud profiling data. Center swath pixels for the same parameters can also be shown as line plots overlaying the Cloudsat or Calipso profile images. Images and subsetted data produced in each analysis run may be downloaded. Users truly can explore and discover data specific to their needs prior to ever transferring data to their analysis tools.

Kempler, S.↗

Software-Hardware Co-design of Heterogeneous SmartNIC System for Recommendation Models Inference and Training

Deep Learning Recommendation Models (DLRMs) are critical applications in various domains and have evolved as one of the single largest machine learning applications. Trillions of DLRM parameters exceed the on-chip memory capacity of GPUs. Large-scale multi-node systems are required for distributed DLRM inference and training, which suffer from the all-to-all communication bottleneck, mainly limiting the scalability of ever-growing DLRMs. In recent years, SmartNICs have evolved with coupled computation and communication capabilities providing opportunities for a powerful heterogeneous device in the system. However, there isn't such a distributed system that fully leverages the abundant smartNIC resources that resolve the scalability issue of DLRMs. In this work, we proposed a software-hardware co-design of a heterogeneous smartNIC system that resolves the communication bottleneck of distributed DLRMs, mitigates the memory bandwidth pressure, and improves computation efficiency. We provide a set of smartNIC designs of cache systems (including local cache and remote cache) and smartNIC computation kernels which reduce data movement, relieve memory lookup intensity, and improve the GPU's computation efficiency. In addition, we propose a graph algorithm that improves the data locality of queries within batches which optimizes the overall system performance with higher data reuse. Our evaluation shows that our system achieves 2.1x latency speedup for inference and 1.6x throughput speedup for training.

Guo, Anqi↗

Understanding the Scalability of Bayesian Network Inference Using Clique Tree Growth Curves

One of the main approaches to performing computation in Bayesian networks (BNs) is clique tree clustering and propagation. The clique tree approach consists of propagation in a clique tree compiled from a Bayesian network, and while it was introduced in the 1980s, there is still a lack of understanding of how clique tree computation time depends on variations in BN size and structure. In this article, we improve this understanding by developing an approach to characterizing clique tree growth as a function of parameters that can be computed in polynomial time from BNs, specifically: (i) the ratio of the number of a BN s non-root nodes to the number of root nodes, and (ii) the expected number of moral edges in their moral graphs. Analytically, we partition the set of cliques in a clique tree into different sets, and introduce a growth curve for the total size of each set. For the special case of bipartite BNs, there are two sets and two growth curves, a mixed clique growth curve and a root clique growth curve. In experiments, where random bipartite BNs generated using the BPART algorithm are studied, we systematically increase the out-degree of the root nodes in bipartite Bayesian networks, by increasing the number of leaf nodes. Surprisingly, root clique growth is well-approximated by Gompertz growth curves, an S-shaped family of curves that has previously been used to describe growth processes in biology, medicine, and neuroscience. We believe that this research improves the understanding of the scaling behavior of clique tree clustering for a certain class of Bayesian networks; presents an aid for trade-off studies of clique tree clustering using growth curves; and ultimately provides a foundation for benchmarking and developing improved BN inference and machine learning algorithms.

Mengshoel, Ole J.↗