Search NASA⌕ Search

SEARCH · Search NASA

Results for “Network inference”

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 1,027 records · Page 57

Meteoroid environment near the Earth-Moon system

We undertook a research program to investigate the properties of small objects crossing the orbit of the Earth-Moon system using a unique set of data obtained during the Apollo lunar landing missions with a network of seismic stations on the surface of the Moon. The primary objectives of the study were to find out the nature of these objects, whether they were of cometary or asteroidal origin based on their orbital distributions and seismic effects upon impacts, and then to infer the role these small objects play in the evolution of the solar system. In this final technical report, we briefly summarize the results of this study. Detailed results of the study have been published in a series of papers and a dissertation, which are listed in the PUBLICATIONS section. Abstracts of the published papers and the dissertation are attached as an appendix.

Nakamura, Yosio↗

Nutrient cycling for biomass: Interactive proteomic/transcriptomic networks for global carbon management processes within poplar-mycorrhizal interactions

This project addresses the need to develop system-scale models at the symbiotic interface between ectomycorrhizal fungi (Laccaria bicolor) and tree species (Populus tremuloides) in response to environmental nutrient availability / biochemistry. Using our now well-established laboratory Laccaria x poplar system, we address the hypothesis that essential regulatory and metabolic mechanisms can be inferred from genomic, transcriptomic and proteomic-level changes that occur in response to environmental nutrient availability. The project addresses this hypothesis by applying state-of-the-art protein-level analytic approaches to fill the gap in our understanding of how mycorrhizal regulatory and metabolic processes at the transcript-level translate to nutrient uptake, carbon management and ultimate net primary productivity of plants. In most cases, these techniques were not previously optimized for poplar trees or Laccaria. Thus, one of the major contributions of this project has been to provide avenues for new research in these species by overcoming the pitfalls that had previously prevented the use of techniques such as ChIP-Seq and SWATH-proteomics. Since it is the proteins that sense and interact with the environment, participate in signal cascades, activate and regulate gene expression, perform the activities of metabolism and ultimately sequester carbon and generate biomass, an understanding of protein activities during symbiosis-linked nutrient uptake is critical to any systems-level approach that links metabolic processes to the environment. This project uses a team of experts at The University of Alabama in Huntsville (UAH), The University of Alabama at Birmingham (UAB) and Argonne National Laboratory (ANL) to address the above hypothesis using a multiple "omics" approach that combines gene and protein expression as well as protein modifications, and biochemical analyses (performed at Brookhaven National Laboratory (BNL)) in poplar trees under mycorrhizal and free living conditions. Together, the assembled team of experts completed all of the planned milestones set forth in this project. In addition to the planned approaches, several lines of exciting new research have also evolved during the course of this project that involved FTIR Imaging using the National Synchrotron Light Source at BNL. A summary of the approaches used in this project and key highlights are as follows: Having the right combination of microbes associated with plants is largely responsible for the plant’s ability to mine nutrients from the soil and to develop a strong “immune system”. Our current chemically focused and intensive culture tends to forget that plants obtain nutrients in two ways: (1) via water soluble chemical nutrients and (2) via the activity of acquired microbial symbionts. In healthy natural ecosystems, chemical nutrients are always in low abundance because the organisms within that system have already locked such nutrients away within the biological system itself. Thus, in nature it is the biological sources of nutrients and the microbes that have the capacity to mine those nutrients for their plant hosts that actually control the terrestrial nutrient cycles on this planet. Thus, a new push in the future may very well be to use our skills at elucidating complex patterns to strategically guide soil microbe communities to do what we want, essentially allowing nature to do the work of figuring out what is most efficient and effective for human needs. However, the findings of this project and other work in our lab lead to the hypothesis that the specific soil community composition is less important than the emergent properties of those communities. So, additional research into what soil communities are effective and how they are established will be key in developing human understanding of how to manipulate biological systems to meet human needs without causing undue damage to our environment.

59 BASIC BIOLOGICAL SCIENCES↗

Combined use of Satellite and Surface Observations to Infer the Imaginary Part of Refractive Index of Saharan Dust

We present a method for retrieval of imaginary part of refractive index of desert dust aerosol in UV part of spectrum along with aerosol layer height above the ground. The method uses Total Ozone Mapping Spectrometer' (TOMS) measurements of the top of atmosphere radiances (331 nm, 360 nm) and aerosol optical depth provided by Aerosol Robotic Network (AERONET) (440 nm). Obtained values of imaginary part of refractive index retrieved for Saharan dust aerosol at 360 nm are significantly lower than previously reported values. The average retrieved values vary between 0.0054 and 0.0066 for different geographical locations. Our findings are in good agreement with the results of several recent investigations. The time variability of retrieved values for aerosol layer height is consistent with the predictions of dust transport model.

Sinyuk, Alexander↗

A resolution independent neural operator

The Deep operator network (DeepONet) is a powerful yet simple neural operator architecture that utilizes two deep neural networks to learn mappings between infinite-dimensional function spaces. This architecture is highly flexible, allowing the evaluation of the solution field at any location within the desired domain. However, it imposes a strict constraint on the input space, requiring all input functions to be discretized at the same locations; this limits its practical applications. Here, in this work, we introduce a general framework for operator learning from input–output data with arbitrary number and locations of sensors. This begins by introducing a resolution-independent DeepONet (RI-DeepONet), enabling it to handle input functions that are arbitrarily, but sufficiently finely, discretized. To this end, we propose two dictionary learning algorithms to adaptively learn a set of appropriate continuous basis functions, parameterized as implicit neural representations (INRs), from correlated signals defined on arbitrary point cloud data. These basis functions are then used to project arbitrary input function data as a point cloud onto an embedding space (i.e., a vector space of finite dimensions) with dimensionality equal to the dictionary size, which can be directly used by DeepONet without any architectural changes. In particular, we utilize sinusoidal representation networks (SIRENs) as trainable INR basis functions. The introduced dictionary learning algorithms are then used in a similar way to learn an appropriate dictionary of basis functions for the output function data, which defines a new neural operator architecture referred to as the R esolution I ndependent N eural O perator (RINO). In the RINO, the operator learning task simplifies to learning a mapping from the coefficients of input basis functions to the coefficients of output basis functions. We demonstrate the robustness and applicability of RINO in handling arbitrarily (but sufficiently richly) sampled input and output functions during both training and inference through several numerical examples.

Deep operator network (DeepONet)↗

Vehicular Re-Identification from Uncontrolled Multiple Views

Vehicle re-identification (re-ID) across disparate sensing modalities remains a fundamental challenge for transportation research. In this work, we introduce a deep multi-view vehicle re-ID framework that leverages Siamese networks to compare pairs of vehicle images and produce matching scores, enabling robust association across drastically different viewpoints such as those from UAVs, surveillance cameras, and ground sensors. The model exploits convolutional neural networks to learn features that remain discriminative under changes in angle, distance, and illumination, supporting more generalizable re-ID performance. As part of this effort, we also developed an automated pipeline to synchronize roadside and UAV video streams, producing a multi-perspective dataset that complements preexisting real collections and a synthetic dataset generated in this study. Together, these contributions advance the capability to re-identify vehicles across wide viewing baselines; establish a foundation for scalable, reproducible research in vehicle re-ID; and open pathways for future applications, such as inferring routine behaviors, movement patterns, and daily habits of the individual associated with the vehicle.

convolutional neural networks↗

Hybrid Recurrent Neural Network Modeling for Traffic Delay Prediction at Signalized Intersections Along an Urban Arterial

This paper studies the traffic delay prediction modeling for multiple signalized intersections along the Ala Moana Boulevard and Nimitz Highway in Hawaii. Several machine learning (ML) based approaches have been studied in the literature, and most of them focused on prediction accuracy rather than the end use of real-time control and implementation. These ML models tend to be very complex and non-linear in nature, making it challenging to achieve fast inferences and are computationally heavy for real-time signal control implementation. As such, in this paper, a simple yet accurate hybrid modeling method is proposed to predict traffic delay one-step ahead with the model made suitable for real-time implementation to control traffic flow. Since real-time road-side measurements are recorded in unstructured form, the paper also discusses other issues related to data extraction and the pre-processing process. Finally, a simple signal control loop is developed to demonstrate the proposed modeling approach, which has shown advantages in model accuracy and computation efficiency compared against several existing modeling methods.

42 ENGINEERING↗

Hypergraph Models of Biological Networks to Identify Genes Critical to Pathogenic Viral Response

Motivation: Representing biological networks as graphs is a powerful approach to reveal underlying patterns, signatures, and critical components from high-throughput biomolecular data. However, graphs do not natively capture the multi-way relationships present among genes and proteins in biological systems such as protein complexes, metabolic reactions, and signal transduction pathways. Hypergraphs are generalizations of graphs that naturally model multi-way interactions in data, and we therefore seek to understand how they can more faithfully identify, and potentially predict, complex relationships in genomic expression data sets. Results: We compiled a novel data set of transcriptional host response to pathogenic viral infections and formulated relationships between genes as a hypergraph where hyperedges are differentially expressed genes and vertices represent conditions. We find that hypergraph betweenness centrality is a superior method for identification of genes important to viral response when compared with graph centrality. Our results demonstrate the utility of using hypergraphs to represent complex biological systems, and highlight potentially interesting biological results about host response to highly pathogenic viruses.

systems biology, hypergraph, viral infection, biol↗

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↗

Deducing neutron star equation of state from telescope spectra with machine-learning-derived likelihoods

The interiors of neutron stars reach densities and temperatures beyond the limits of terrestrial experiments, providing vital laboratories for probing nuclear physics. While the star's interior is not directly observable, its pressure and density determine the star's macroscopic structure which affects the spectra observed in telescopes. The relationship between the observations and the internal state is complex and partially intractable, presenting difficulties for inference. Previous work has focused on the regression from stellar spectra of parameters describing the internal state. We demonstrate a calculation of the full likelihood of the internal state parameters given observations, accomplished by replacing intractable elements with machine learning models trained on samples of simulated stars. Our machine-learning-derived likelihood allows us to perform maximum a posteriori estimation of the parameters of interest, as well as full scans. We demonstrate the technique by inferring stellar mass and radius from an individual stellar spectrum, as well as equation of state parameters from a set of spectra. Our results are more precise than pure regression models, reducing the width of the parameter residuals by 11.8% in the most realistic scenario. The neural networks will be released as a tool for fast simulation of neutron star properties and observed spectra.

79 ASTRONOMY AND ASTROPHYSICS↗

Cost‐effective monitoring of biological invasions under global change: a model‐based framework

Summary Ecological monitoring programmes are designed to detect and measure changes in biodiversity and ecosystems. In the case of biological invasions, they can contribute to anticipating risks and adaptively managing invaders. However, monitoring is often expensive because large amounts of data might be needed to draw inferences. Thus, careful planning is required to ensure that monitoring goals are realistically achieved. Species distribution models ( SDM s) can provide estimates of suitable areas to invasion. Predictions from these models can be applied as inputs in optimization strategies seeking to identify the optimal extent of the networks of areas required for monitoring risk of invasion under current and future environmental conditions. A hierarchical framework is proposed herein that combines SDM s, scenario analysis and cost analyses to improve invasion assessments at regional and local scales. We illustrate the framework with Acacia dealbata Link. (Silver‐wattle) in northern Portugal. The framework is general and applicable to any species. We defined two types of monitoring networks focusing either on the regional‐scale management of an invasion, or management focus within and around protected areas. For each one of these two schemes, we designed a hierarchical framework of spatial prioritization using different information layers (e.g. SDM s, habitat connectivity, protected areas). We compared the performance of each monitoring scheme against 100 randomly generated models. In our case study, we found that protected areas will be increasingly exposed to invasion by A. dealbata due to climate change. Moreover, connectivity between suitable areas for A. dealbata is predicted to increase. Monitoring networks that we identify were more effective in detecting new invasions and less costly to management than randomly generated models. The most cost‐efficient monitoring schemes require 18% less effort than the average networks across all of the 100 tested options. Synthesis and applications . The proposed framework achieves cost‐effective monitoring networks, enabling the interactive exploration of different solutions and the combination of quantitative information on network performance with orientations that are rarely incorporated in a decision support system. The framework brings invasion monitoring closer to European legislation and management needs while ensuring adaptability under rapid climate and environmental change.

Vicente, Joana R.↗

Localization Leads to Improved Distributed Detection Under Non-Smooth Distributions

We consider a detection network of sensors that measure intensity levels due to a source amidst background inside a two-dimensional monitoring area. The source intensity decays away from it possibly in discrete jumps, and the corresponding sensor measurements could be random due to the nature of source and background, or due to sensor errors, or both. The detection problem is to infer the presence of a source based on sensor measurements. In the conventional decision/detection fusion approach, detection decisions are made at the individual sensors using Sequential Probability Ratio Test (SPRT), and are combined at the fusion center using a Boolean fusion rule. We show that better detection can be achieved by utilizing sensor measurements at the fusion center, by first localizing the source and then utilizing a more effective SPRT. This approach leads to the detection performance superior to any Boolean detection fuser, under fairly general conditions: (i) smooth and non-smooth source intensity functions and probability ratios, and (ii) a minimum packing number of the state-space. We apply this method to improve the detection of (a) low-level point radiation sources amidst background radiation under strong shielding conditions, and (b) the well-studied Gaussian source amidst Gaussian background.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Learning functional priors and posteriors from data and physics

In this work, we develop a new Bayesian framework based on deep neural networks to be able to extrapolate in space-time using historical data and to quantify uncertainties arising from both noisy and gappy data in physical problems. Specifically, the proposed approach has two stages: (1) prior learning and (2) posterior estimation. At the first stage, we employ the physics-informed Generative Adversarial Networks (PI-GAN) to learn a functional prior either from a prescribed function distribution, e.g., Gaussian process, or from historical data and physics. At the second stage, we employ the Hamiltonian Monte Carlo (HMC) method to estimate the posterior in the latent space of PI-GANs. In addition, we use two different approaches to encode the physics: (1) automatic differentiation, used in the physicsinformed neural networks (PINNs) for scenarios with explicitly known partial differential equations (PDEs), and (2) operator regression using the deep operator network (DeepONet) for PDE-agnostic scenarios. We then test the proposed method for (1) meta-learning for one-dimensional regression, and forward/inverse PDE problems (combined with PINNs); (2) PDE-agnostic physical problems (combined with DeepONet), e.g., fractional diffusion as well as saturated stochastic (100-dimensional) flows in heterogeneous porous media; and (3) spatial-temporal regression problems, i.e., inference of a marine riser displacement field using experimental data from the Norwegian Deepwater Programme (NDP). The results demonstrate that the proposed approach can provide accurate predictions as well as uncertainty quantification given very limited scattered and noisy data, since historical data could be available to provide informative priors. In summary, the proposed method is capable of learning flexible functional priors, e.g., both Gaussian and non-Gaussian process, and can be readily extended to big data problems by enabling mini-batch training using stochastic HMC or normalizing flows since the latent space is generally characterized as low dimensional.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

An expanded transcriptome atlas for Bacteroides thetaiotaomicron reveals a small RNA that modulates tetracycline sensitivity

Plasticity in gene expression allows bacteria to adapt to diverse environments. This is particularly relevant in the dynamic niche of the human intestinal tract; however, transcriptional networks remain largely unknown for gut-resident bacteria. Here we apply differential RNA sequencing (RNA-seq) and conventional RNA-seq to the model gut bacterium Bacteroides thetaiotaomicron to map transcriptional units and profile their expression levels across 15 in vivo-relevant growth conditions. We infer stress- and carbon source-specific transcriptional regulons and expand the annotation of small RNAs (sRNAs). Integrating this expression atlas with published transposon mutant fitness data, we predict conditionally important sRNAs. These include MasB, which downregulates tetracycline tolerance. Using MS2 affinity purification and RNA-seq, we identify a putative MasB target and assess its role in the context of the MasB-associated phenotype. These data—publicly available through the Theta-Base web browser (http://micromix.helmholtz-hiri.de/bacteroides/)—constitute a valuable resource for the microbiome community.

59 BASIC BIOLOGICAL SCIENCES↗

Nonlinear intrinsic variables and state reconstruction in multiscale simulations

Finding informative low-dimensional descriptions of high-dimensional simulation data (like the ones arising in molecular dynamics or kinetic Monte Carlo simulations of physical and chemical processes) is crucial to understanding physical phenomena, and can also dramatically assist in accelerating the simulations themselves. In this paper, we discuss and illustrate the use of nonlinear intrinsic variables (NIV) in the mining of high-dimensional multiscale simulation data. In particular, we focus on the way NIV allows us to functionally merge different simulation ensembles, and different partial observations of these ensembles, as well as to infer variables not explicitly measured. The approach relies on certain simple features of the underlying process variability to filter out measurement noise and systematically recover a unique reference coordinate frame. We illustrate the approach through two distinct sets of atomistic simulations: a stochastic simulation of an enzyme reaction network exhibiting both fast and slow time scales, and a molecular dynamics simulation of alanine dipeptide in explicit water.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Linewidth broadening and tunneling of excitons bound to N pairs in dilute GaAs:N

The exciton bound to a pair of nitrogen atoms situated at nearby lattice sites in dilute GaAs:N provides an energetically uniform electronic system, spectrally distinct from pairs with larger or smaller separations, and can even be grown with a uniform pair orientation in the crystal. We use photoluminescence excitation spectroscopy on an ensemble of N pairs to study the narrow continuous energy distribution within two of the individual exchange- and symmetry-split exciton states. Inhomogeneous linewidths of 50-60 ueV vary across the crystal on a mesoscopic scale and can be 30 ueV at microscopic locations indicating that the homogeneous linewidth inferred from previous time-domain measurements is still considerably broadened. While excitation and emission linewidths are similar, results show a small energy shift between them indicative of exciton transfer via phonon-assisted tunneling between spatially separated N pairs. We numerically simulate the tunneling in a spatial network of randomly distributed pairs having a normal distribution of bound exciton energies. Comparing the ensemble excitation-emission energy shift with the measured results shows that the transfer probability is higher than expected from the dilute pair concentration and what is known of the exciton wavefunction spatial extent. Both the broadening and the exciton transfer have implications for the goal of pair-bound excitons as a single- or multi-qubit system.

36 MATERIALS SCIENCE↗

A hybrid neural architecture: Online attosecond x-ray characterization

The emergence of high-repetition-rate x-ray free-electron lasers (XFELs), such as SLAC’s LCLS-II, serves as our canonical example for autonomous controls that necessitate high-throughput diagnostics paired with streaming computational pipelines capable of single-shot analysis with extremely low latency. We present the deterministic characterization with an integrated parallelizable hybrid resolver architecture, a hybrid machine learning framework designed for fast, accurate analysis of XFEL diagnostics using angular streaking-based sinogram images. This architecture integrates convolutional neural networks and bidirectional long short-term memory models to denoise input, identify x-ray sub-spike features, and extract sub-spike relative delays with sub-30 attosecond temporal resolution. Deployed on low-latency hardware, it achieves over 10 kHz throughput with 168.3 μs inference latency, indicating scalability to 14 kHz with field-programmable gate array integration. By transforming regression tasks into classification problems and leveraging optimized error encoding, we achieve high precision with low-latency performance that is critical for real-time streaming event selection and experimental control feedback signals. This represents a key development in real-time control pipelines for next-generation autonomous science, generally, and high repetition-rate x-ray experiments in particular.

Accelerator Physics (physics.acc-ph)↗

Towards Improved Real-Time Observability of Behind-Meter PhotoVoltaic Systems: A Data-Driven Approach

The stochasticity in energy supply introduced by increased PV penetration in grid distribution systems can lead to frequent and sudden supply-demand imbalances, thereby causing voltage and frequency fluctuations. Enhanced observability, in real-time, of grid attributes such as real and reactive power flows, power injections/consumption and voltages is becoming increasingly essential to ensure smooth grid operation under these new large-scale PV integration scenarios. Two significant challenges prevent grid operators from achieving this objective: 1) Computational complexity of quasi-static timeseries simulations, and 2) Partial observability of behind-meter grid assets. In this work, we take a step towards this objective by enhancing the observability of smart inverter equipped PV systems located behind AMI meters. Smart inverters convert the DC output of PV systems into real and reactive power, the ratio of which is governed by a power factor setting. This in turn is governed by the real-time voltage at its terminal, as per a pre-set volt/var curve. While it is possible to infer the power factor setting - thereby, real and reactive power injection of the PV system - at each time interval by quasi-static time series simulations, such methods are not scalable to large grid sizes. We propose to reduce the reliance on simulations by developing a deep neural network based data-driven model which predicts the time dependent power factors of the PV systems in the grid. The model uses a small amount of simulated data to learn the dependence of power factor on 1) load and apparent PV power injection (root mean square of real and reactive power) data, and 2) AMI net-load data (for BTM PV assets). Our results show that our model can predict power factors with mean absolute error of less than 6.2% under full observability scenario and under 8.6% under the partial observability scenario. Thus, our data-driven model can reduce the amount of simulations required to compute real and reactive injections from PV systems in the distribution system with or without complete information.

14 SOLAR ENERGY↗

3D-ReG: A 3D ReRAM-based Heterogeneous Architecture for Training Deep Neural Networks

Deep neural network (DNN) models are being expanded to a broader range of applications. The computational capability of traditional hardware platforms cannot accommodate the growth of model complexity. Among recent technologies to accelerate DNN, resistive memory (ReRAM)-based processing-in-memory (PIM) emerged as a promising solution for DNN inference due to its high efficiency for matrix-based computation. We face two major technical challenges in extending the use of ReRAM-based accelerators for training: (1) full-precision data is essential in back-propagation; (2) the need to support both feed-forward and back-propagation aggravates the data-movement burden. We propose a heterogeneous architecture named as 3D-ReG, which leverages full-precision GPU to ensure training accuracy and low-overhead 3D integration to provide low-cost data movements. Moreover, we introduce conservative and aggressive task-mapping schemes, which partition the computation phases in different ways to balance execution efficiency and training accuracy. We evaluate 3D-ReG implemented with two 3D integration technologies, through-silicon vias (TSVs) and monolithic inter-tier vias (MIVs), and compare them with GPU-only and PIM-only counterparts. Various GPU-only platforms using two main-memory technologies (DRAM, ReRAM) and three interconnect technologies (2D, TSV, MIV) are evaluated as well. Experimental results show that 3D-ReG can achieve on average 5.64× training speedup and 3.56× higher energy efficiency compared with the GPU with DRAM as main memory, at the cost of 0.05%–3.39% accuracy drop. We define a new metric, gain-loss ratio (GLR), which quantitatively evaluates the capability of a DNN training hardware in terms of the model accuracy and hardware efficiency. The results of our comparison show that the aggressive task-mapping scheme on MIV-based 3D-ReG outperforms the other methods.

Computer Science↗