Search NASA⌕ Search

SEARCH · Search NASA

Results for “imbalanced data”

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 73 records · Page 4

On Neural Architectures for Astronomical Time-series Classification with Application to Variable Stars

Despite the utility of neural networks (NNs) for astronomical time-series classification, the proliferation of learning architectures applied to diverse data sets has thus far hampered a direct intercomparison of different approaches. Here we perform the first comprehensive study of variants of NN-based learning and inference for astronomical time series, aiming to provide the community with an overview on relative performance and, hopefully, a set of best-in-class choices for practical implementations. In both supervised and self-supervised contexts, we study the effects of different time-series-compatible layer choices, namely the dilated temporal convolutional neural network (dTCNs), long-short term memory NNs, gated recurrent units and temporal convolutional NNs (tCNNs). Additionally, we also study the efficacy and performance of encoder-decoder (i.e., autoencoder) networks compared to direct classification networks, different pathways to include auxiliary (non-time-series) metadata, and different approaches to incorporate multi-passband data (i.e., multiple time series per source). Performance—applied to a sample of 17,604 variable stars (VSs) from the MAssive Compact Halo Objects (MACHO) survey across 10 imbalanced classes—is measured in training convergence time, classification accuracy, reconstruction error, and generated latent variables. We find that networks with recurrent NNs generally outperform dTCNs and, in many scenarios, yield to similar accuracy as tCNNs. In learning time and memory requirements, convolution-based layers perform better. We conclude by discussing the advantages and limitations of deep architectures for VS classification, with a particular eye toward next-generation surveys such as the Legacy Survey of Space and Time, the Roman Space Telescope, and Zwicky Transient Facility.

79 ASTRONOMY AND ASTROPHYSICS↗

Using convolutional neural networks to detect edge localized modes in DIII-D from Doppler backscattering measurements

In H-mode tokamak plasmas, the plasma is sometimes ejected beyond the edge transport barrier. These events are known as edge localized modes (ELMs). ELMs cause a loss of energy and damage the vessel walls. Understanding the physics of ELMs, and by extension, how to detect and mitigate them, is an important challenge. In this paper, we focus on two diagnostic methods—deuterium-alpha (D α ) spectroscopy and Doppler backscattering (DBS). The former detects ELMs by measuring Balmer alpha emission, while the latter uses microwave radiation to probe the plasma. DBS has the advantages of having a higher temporal resolution and robustness to damage. These advantages of DBS diagnostic may be beneficial for future operational tokamaks, and thus, data processing techniques for DBS should be developed in preparation. In sight of this, we explore the training of neural networks to detect ELMs from DBS data, using D α data as the ground truth. With shots found in the DIII-D database, the model is trained to classify each time step based on the occurrence of an ELM event. The results are promising. When tested on shots similar to those used for training, the model is capable of consistently achieving a high f1-score of 0.93. Furthermore, this score is a performance metric for imbalanced datasets that ranges between 0 and 1. We evaluate the performance of our neural network on a variety of ELMs in different high confinement regimes (grassy ELM, RMP mitigated, and wide-pedestal), finding broad applicability. Beyond ELMs, our work demonstrates the wider feasibility of applying neural networks to data from DBS diagnostic.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Thermodynamics of balanced and slightly spin-imbalanced Fermi gases at unitarity

In this article we present a Monte Carlo calculation of the critical temperature and other thermodynamic quantities for the unitary Fermi gas with a population imbalance (unequal number of fermions in the two spin components). We describe an improved worm-type algorithm that is less prone to autocorrelations than the previously available methods and show how this algorithm can be applied to simulate the unitary Fermi gas in presence of a small imbalance. Our data indicate that the critical temperature remains almost constant for small imbalances h={Delta}{mu}/{epsilon}{sub F} < or approx. 0.2. We obtain the continuum result T{sub c}=0.171(5){epsilon}{sub F} in units of Fermi energy and derive a lower bound on the deviation of the critical temperature from the balanced limit, T{sub c}(h)-T{sub c}(0)>-0.5{epsilon}{sub F}h{sup 2}. Using an additional assumption a tighter lower bound can be obtained. We also calculate the energy per particle and the chemical potential in the balanced and imbalanced cases.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A comparison of machine learning methods to classify radioactive elements using prompt-gamma-ray neutron activation data

The detection of illicit radiological materials is critical to establishing a robust second line of defence in nuclear security. Neutron-capture prompt-gamma activation analysis (PGAA) can be used to detect multiple radioactive materials across the entire Periodic Table. However, long detection times and a high rate of false positives pose a significant hindrance in the deployment of PGAA-based systems to identify the presence of illicit substances in nuclear forensics. In the present work, six different machine-learning algorithms were developed to classify radioactive elements based on the PGAA energy spectra. The model performance was evaluated using standard classification metrics and trend curves with an emphasis on comparing the effectiveness of algorithms that are best suited for classifying imbalanced datasets. We analyse the classification performance based on Precision, Recall, F1-score, Specificity, Confusion matrix, ROC-AUC curves, and Geometric Mean Score (GMS) measures. The tree-based algorithms (Decision Trees, Random Forest and AdaBoost) have consistently outperformed Support Vector Machine and K-Nearest Neighbours. Based on the results presented, AdaBoost is the preferred classifier to analyse data containing PGAA spectral information due to the high recall and minimal false negatives reported in the minority class.

97 MATHEMATICS AND COMPUTING↗

AWB-GCN: A Graph Convolutional Network Accelerator with Runtime Workload Rebalancing

The recent development of deep learning has been mostly focusing on Euclidean data, such as images, videos, audios, etc. However, most real-world information and relation are often expressed as graphs. To efficiently learn from graph data, graph convolutional networks (GCNs) emerge as a promising approach, showing advantages in several practical applications such as social network analysis, knowledge discovery, 3D modeling, motion capturing, etc. Real-world graphs are usually extremely large and imbalanced, posting significant performance demand and design challenges on the hardware dedicated for GCN inference. In this paper, we propose an architecture design called UW-GCN to accelerate graph convolutional network inference. To tackle the major performance bottleneck from workload imbalance, we propose dynamic neighborhood stealing and remote chunk shuffling techniques, relying on hardware flexibility to achieve hardware auto-tuning under negligible area or delay overhead. Specifically, UW-GCN is able to smartly profile the sparse graph pattern while continuously adjusting the workload distribution via routing reconfiguration among parallel processing elements (PEs). The ideal configuration is then reused in the remaining iterations. To the best of our knowledge, this is the first accelerator design particularly for GCN and the first work relying on hardware auto-tuning, which is normally based on software, to achieve near-optimal workload balance in processing sparse structures.

Geng, Tong↗

Reducing the Parameter Dependency of Phase-Picking Neural Networks with Dice Loss

Training a neural network for picking seismic phase arrivals has been commonly posed as a segmentation problem. It is a highly imbalanced segmentation problem in the sense that the background vastly dominates the foreground because we are trying to pick the optimal single sample point that represents the arrival of a seismic phase in a many seconds long time window. Here, we test the Dice loss, which is a preferred loss function for highly imbalanced image segmentation problems. We show that phase-picking neural networks trained on the Dice loss behave in a binary fashion for which the prediction output is almost always either nearly 1 or nearly 0. This feature removes the strong dependence of data processing workflows on the prediction score threshold, which is an otherwise critical parameter to determine when using neural networks trained on the cross-entropy loss. When strategically used, models trained on the Dice loss can reduce the parameter dependency of machine learning-based seismic monitoring.

58 GEOSCIENCES↗

DART: distributed adaptive radix tree for efficient affix-based keyword search on HPC systems

© 2018 Association for Computing Machinery. Affix-based search is a fundamental functionality for storage systems. It allows users to find desired datasets, where attributes of a dataset match an affix. While building inverted index to facilitate efficient affix based keyword search is a common practice for standalone databases and for desktop file systems, building local indexes or adopting indexing techniques used in a standalone data store is insufficient for highperformance computing (HPC) systems due to the massive amount of data and distributed nature of the storage devices within a system. In this paper, we propose Distributed Adaptive Radix Tree (DART), to address the challenge of distributed affix-based keyword search on HPC systems. This trie-based approach is scalable in achieving efficient affix-based search and alleviating imbalanced keyword distribution and excessive requests on keywords at scale. Our evaluation at different scales shows that, comparing with the "full string hashing" use case of the most popular distributed indexing technique - Distributed Hash Table (DHT), DART achieves up to 55× better throughput with prefix search and with suffix search, while achieving comparable throughput with exact and infix searches. Also, comparing to the "initial hashing" use case of DHT, DART maintains a balanced keyword distribution on distributed nodes and alleviates excessive query workload against popular keywords.

Zhang, Wei↗

An asynchronous traversal engine for graph-based rich metadata management

Rich metadata in high-performance computing (HPC) systems contains extended information about users, jobs, data files, and their relationships. Property graphs are a promising data model to represent heterogeneous rich metadata flexibly. Specifically, a property graph can use vertices to represent different entities and edges to record the relationships between vertices with unique annotations. The high-volume HPC use case, with millions of entities and relationships, naturally requires an out-of-core distributed property graph database, which must support live updates (to ingest production information in real time), low-latency point queries (for frequent metadata operations such as permission checking), and large-scale traversals (for provenance data mining). Among these needs, large-scale property graph traversals are particularly challenging for distributed graph storage systems. Most existing graph systems implement a "level synchronous" breadth-first search algorithm that relies on global synchronization in each traversal step. This performs well in many problem domains; but a rich metadata management system is characterized by imbalanced graphs, long traversal lengths, and concurrent workloads, each of which has the potential to introduce or exacerbate stragglers (i.e., abnormally slow steps or servers in a graph traversal) that lead to low overall throughput for synchronous traversal algorithms. Previous research indicated that the straggler problem can be mitigated by using asynchronous traversal algorithms, and many graph-processing frameworks have successfully demonstrated this approach. Such systems require the graph to be loaded into a separate batch-processing framework instead of being iteratively accessed, however. In this work, we investigate a general asynchronous graph traversal engine that can operate atop a rich metadata graph in its native format. We outline a traversal-aware query language and key optimizations (traversal-affiliate caching and execution merging) necessary for efficient performance. We further explore the effect of different graph partitioning strategies on the traversal performance for both synchronous and asynchronous traversal engines. Our experiments show that the asynchronous graph traversal engine is more efficient than its synchronous counterpart in the case of HPC rich metadata processing, where more servers are involved and larger traversals are needed. Furthermore, the asynchronous traversal engine is more adaptive to different graph partitioning strategies.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Skew-Symmetric adjacency matrices for clustering directed graphs

Graph clustering methods often critically rely on the symmetry of graph matrices. Developing analogous methods for digraphs often proves more challenging, because digraph matrices are typically asymmetric and not orthogonally diagonalizable. However, researchers have recently proposed several complex-valued Hermitian digraph matrices. In particular, one such representation has been utilized as an input to an algorithm for finding imbalanced cuts. In this work, we establish an algebraic relationship between this matrix and an associated real-valued matrix. We show using this real-valued matrix for imbalanced cut-finding algorithms is not only sufficient but advantageous. Our algorithm uses less memory and asymptotically less computation while provably preserving solution quality. We also show our method can be easily implemented using standing computational building blocks, possesses better numerical properties, and loans itself to a natural interpretation via an objective function relaxation argument. We empirically demonstrate these advantages on real world data sets and show how our algorithm can uncover meaningful cluster structure.

Hayashi, Koby↗

Oxygenation influences xylose fermentation and gene expression in the yeast genera Spathaspora and Scheffersomyces

Abstract Background Cost-effective production of biofuels from lignocellulose requires the fermentation of d -xylose. Many yeast species within and closely related to the genera Spathaspora and Scheffersomyces (both of the order Serinales) natively assimilate and ferment xylose. Other species consume xylose inefficiently, leading to extracellular accumulation of xylitol. Xylitol excretion is thought to be due to the different cofactor requirements of the first two steps of xylose metabolism. Xylose reductase (XR) generally uses NADPH to reduce xylose to xylitol, while xylitol dehydrogenase (XDH) generally uses NAD + to oxidize xylitol to xylulose, creating an imbalanced redox pathway. This imbalance is thought to be particularly consequential in hypoxic or anoxic environments. Results We screened the growth of xylose-fermenting yeast species in high and moderate aeration and identified both ethanol producers and xylitol producers. Selected species were further characterized for their XR and XDH cofactor preferences by enzyme assays and gene expression patterns by RNA-Seq. Our data revealed that xylose metabolism is more redox balanced in some species, but it is strongly affected by oxygen levels. Under high aeration, most species switched from ethanol production to xylitol accumulation, despite the availability of ample oxygen to accept electrons from NADH. This switch was followed by decreases in enzyme activity and the expression of genes related to xylose metabolism, suggesting that bottlenecks in xylose fermentation are not always due to cofactor preferences. Finally, we expressed XYL genes from multiple Scheffersomyces species in a strain of Saccharomyces cerevisiae . Recombinant S. cerevisiae expressing XYL1 from Scheffersomyces xylosifermentans , which encodes an XR without a cofactor preference, showed improved anaerobic growth on xylose as the primary carbon source compared to S. cerevisiae strain expressing XYL genes from Scheffersomyces stipitis . Conclusion Collectively, our data do not support the hypothesis that xylitol accumulation occurs primarily due to differences in cofactor preferences between xylose reductase and xylitol dehydrogenase; instead, gene expression plays a major role in response to oxygen levels. We have also identified the yeast Sc. xylosifermentans as a potential source for genes that can be engineered into S. cerevisiae to improve xylose fermentation and biofuel production.

09 BIOMASS FUELS↗

DeepAndes: A Self-Supervised Vision Foundation Model for Multispectral Remote Sensing Imagery of the Andes

By mapping sites at large scales usingremotely sensed data, archaeologists can generate unique insights into long-term demographic trends, interregional social networks, and human adaptations in the past. Remote sensing surveys complement field-based approaches, and their reach can be especially great when combined with deep learning and computer vision techniques. However, conventional supervised deep learning methods face challenges in annotating fine-grained archaeological features at scale. In addition, while recent vision foundation models have shown remarkable success in learning large-scale remote sensing data with minimal annotations, most off-the-shelf solutions are designed for RGB images rather than multispectral satellite imagery, such as the eight-band data used in our study. In this article, we introduce DeepAndes, a transformer-based vision foundation model trained on three million multispectral satellite images, specifically tailored for Andean archaeology. DeepAndes incorporates a customized DINOv2 self-supervised learning algorithm optimized for eight-band multispectral imagery, marking the first foundation model designed explicitly for the Andes region. We evaluate its image understanding performance through imbalanced image classification, image instance retrieval, and pixel-level semantic segmentation tasks. Our experiments show that DeepAndes achieves superior F1 scores, mean average precision, and Dice scores in few-shot learning scenarios, significantly outperforming models trained from scratch or pretrained on smaller datasets. This underscores the effectiveness of large-scale self-supervised pretraining in archaeological remote sensing.

Guo, Junlin [Vanderbilt Univ., Nashville, TN (Unit↗

Measurements of the suppression and correlations of dijets in Pb+Pb collisions at $\sqrt{s_{NN}}$ = 5.02 TeV

Studies of the correlations of the two highest transverse momentum (leading) jets in individual Pb+Pb collision events can provide information about the mechanism of jet quenching by the hot and dense matter created in such collisions. In Pb+Pb and $pp$ collisions at $\sqrt{s_{NN}}$ = 5.02 TeV, measurements of the leading dijet transverse momentum ($p_T$) correlations are presented. Additionally, measurements in Pb+Pb collisions of the dijet pair nuclear modification factors projected along leading and subleading jet $p_T$ are made. The measurements are performed using the ATLAS detector at the LHC with 260 pb –1 of pp data collected in 2017 and 2.2 nb –1 of Pb+Pb data collected in 2015 and 2018. An unfolding procedure is applied to the two-dimensional leading and subleading jet $p_T$ distributions to account for experimental effects in the measurement of both jets. Results are provided for dijets with leading jet $p_T$ greater than 100 GeV. Measurements of the dijet-yield-normalized $x_{J}$ distributions in Pb+Pb collisions show an increased fraction of imbalanced jets compared to pp collisions; these measurements are in agreement with previous measurements of the same quantity at 2.76 TeV in the overlapping kinematic range. Measurements of the absolutely normalized dijet rate in Pb+Pb and pp collisions are also presented, and show that balanced dijets are significantly more suppressed than imbalanced dijets in Pb+Pb collisions. It is observed in the measurements of the pair nuclear modification factors that the subleading jets are significantly suppressed relative to leading jets with $p_T$ between 100 and 316 GeV for all centralities in Pb+Pb collisions.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

New Results on Communication- and Memory-Aware Load Balancing Model and Algorithms

While load balancing in distributed-memory computing has been well-studied, we present an innovative approach to this problem: a unified, reduced-order model that combines three key components to describe “work” in a distributed system: computation, communication, and memory. Our model enables an optimizer to explore complex tradeoffs in task placement, such as augmented parallelism, at the expense of data replication increasing memory usage. We propose a fully distributed, heuristic-based load balancing optimization algorithm, and demonstrate that it quickly finds close-to-optimal solutions. We formalize the complex optimization problem as a mixed-integer linear program, and compare it to our strategy. Finally, we show that when applied to an electromagnetics code, our approach obtains up to 2.3x speedups for the imbalanced execution.

97 MATHEMATICS AND COMPUTING↗

MAGNETOHYDRODYNAMIC WAVES AND CORONAL HEATING: UNIFYING EMPIRICAL AND MHD TURBULENCE MODELS

We present a new global model of the solar corona, including the low corona, the transition region, and the top of the chromosphere. The realistic three-dimensional magnetic field is simulated using the data from the photospheric magnetic field measurements. The distinctive feature of the new model is incorporating MHD Alfven wave turbulence. We assume this turbulence and its nonlinear dissipation to be the only momentum and energy source for heating the coronal plasma and driving the solar wind. The difference between the turbulence dissipation efficiency in coronal holes and that in closed field regions is because the nonlinear cascade rate degrades in strongly anisotropic (imbalanced) turbulence in coronal holes (no inward propagating wave), thus resulting in colder coronal holes, from which the fast solar wind originates. The detailed presentation of the theoretical model is illustrated with the synthetic images for multi-wavelength EUV emission compared with the observations from SDO AIA and STEREO EUVI instruments for the Carrington rotation 2107.

79 ASTRONOMY AND ASTROPHYSICS↗

AIMFAST : an alignment tool based on fringe reflection methods applied to dish concentrators.

The proper alignment of facets on a dish engine concentrated solar power system is critical to the performance of the system. These systems are generally highly concentrating to produce high temperatures for maximum thermal efficiency so there is little tolerance for poor optical alignment. Improper alignment can lead to poor performance and shortened life through excessively high flux on the receiver surfaces, imbalanced power on multicylinder engines, and intercept losses at the aperture. Alignment approaches used in the past are time consuming field operations, typically taking 4-6 h per dish with 40-80 facets on the dish. Production systems of faceted dishes will need rapid, accurate alignment implemented in a fraction of an hour. In this paper, we present an extension to our Sandia Optical Fringe Analysis Slope Technique mirror characterization system that will automatically acquire data, implement an alignment strategy, and provide real-time mirror angle corrections to actuators or labor beneath the dish. The Alignment Implementation for Manufacturing using Fringe Analysis Slope Technique (AIMFAST) has been implemented and tested at the prototype level. In this paper we present the approach used in AIMFAST to rapidly characterize the dish system and provide near-real-time adjustment updates for each facet. The implemented approach can provide adjustment updates every 5 s, suitable for manual or automated adjustment of facets on a dish assembly line.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DETAILED FIT OF 'CRITICAL BALANCE' THEORY TO SOLAR WIND TURBULENCE MEASUREMENTS

We derive the reduced spectrum of turbulent magnetic fluctuations at different frequencies f which would be observed by a single spacecraft in the solar wind when the magnetic field was at an angle {theta}{sub B} to the solar wind flow, if the wavevector spectrum in the solar wind frame were in anisotropic 'critical balance' (CB) as proposed by Goldreich and Sridhar in 1995 (GS95). The anisotropic power spectrum in the inertial range, P(f, {theta}{sub B} ), is scaled onto one curve with f-{sup 5/3} behavior at {theta}{sub B} near 90{sup 0} and f{sup -2} behavior at small {theta}{sub B}. The transition between the two limiting spectra depends on the form of the GS95 wavevector spectrum and the CB scaling parameter L. Using wavelet analysis of Ulysses magnetic field data in three 30-day periods in the high-latitude solar wind in 1995, we verify that the scaling of power with angle and frequency is qualitatively consistent with GS95 theory. However, the scale length L required to fit the observed P(f, {theta}{sub B} ) to the original CB theory is rather less than the scale predicted by that theory for the solar wind. Part, possibly all, of this discrepancy is removed when the GS95 theory modified for imbalanced turbulence is used.

79 ASTRONOMY AND ASTROPHYSICS↗

Upregulation of heme oxygenase-1 expression by dehydrodiconiferyl alcohol (DHCA) through the AMPK–Nrf2 dependent pathway

Oxidative stress is induced by the accumulation of free radicals, resulting in an imbalanced cellular redox state, which has been implicated in a variety of human diseases. Dehydrodiconiferyl alcohol (DHCA), a lignan compound isolated from Cucurbita moschata, has previously been reported to contain anti-adipogenic and anti-lipogenic effects on 3T3-L1 cells and primary MEFs (Abraham and Kappas, 2008). In this study, it was tested whether DHCA could affect the expression of HO-1, using Raw264.7 mouse macrophage cell line. DHCA increased the protein and RNA levels of HO-1 and upregulated its promoter activity. Data from transient transfection assays indicated that ARE located in the E1 region of the HO-1 promoter are important in this DHCA-mediated induction of HO-1 expression. DHCA was also shown to enhance the nuclear translocation and binding of Nrf2 to the respective DNA sequences. The upregulation of HO-1 expression by DHCA was also observed in primary macrophages derived from wild type animals, but not in those from Nrf2 KO mice. Effects of DHCA on HO-1 and Nrf2 were reduced when cells were treated with an AMPK inhibitor, Compound C, but not by PI3K/Akt or MAPK inhibitors. Data from an experiment using a specific siRNA or chemical inhibitor for HO-1 suggested that the DHCA-mediated induction of the HO-1 protein could suppress the LPS-stimulated production of NO. Taken together, our data suggest that DHCA induces the expression of HO-1 by controlling its promoter activity through the AMPK–Nrf2 pathway, eventually leading to the reduction of NO production, and may thus have potential as an effective antioxidant. - Highlights: • Dehydrodiconiferyl alcohol (DHCA) induced the expression of heme oxygenase (HO)-1. • The AMPK–Nrf2 pathway is critically involved in the DHCA-mediated induction of HO-1. • DHCA increased the expression of HO-1, Gclc and Gclm in primary macrophages. • DHCA-mediated induction of HO-1 contributed to the suppression of NO production.

60 APPLIED LIFE SCIENCES↗

Comparative Assessment of U-Net-Based Deep Learning Models for Segmenting Microfractures and Pore Spaces in Digital Rocks

Segmentation of high-resolution X-ray microcomputed tomography (µCT) images is crucial in digital rock physics (DRP), affecting the characterization and analysis of microscale phenomena in the porous media. The complexity of geological structures and nonideal scanning conditions pose significant challenges to conventional image segmentation approaches. Motivated by the recent increasing popularity of deep learning (DL) techniques in image processing, this work undertakes a comparative study of DL models, specifically U-Net and its variants, for segmenting multiple targets with distinguished features in digital rocks, including discrete fracture networks (DFNs), pore spaces, and solid rock. Particularly, DFNs have a smaller volumetric fraction over others, bringing in a substantial challenge of imbalanced segmentation. The primary focus is to evaluate the architecture and feature enhancement strategies of various DL models, including U-Net, attention U-Net, residual U-Net, U-Net++, and residual U-Net++. The models were designed as 2.5D, utilizing a central 2D image and its two adjacent upper and lower 2D images as input to provide a pseudo-3D context. In addition, because the ground truth of segmentation was unknown for real-world digital rocks, we created a benchmark data set following the inverse operations of segmentation. The data synthesis started from the label images (i.e., solid rock, pore spaces, and DFNs), followed by simulating partial volume blurring, adding random background noise, and introducing ring artifacts to mimic real raw X-ray µCT images. The data set, which included various rock types (i.e., sandstone and artificial data), scanning resolution, and magnitudes of noise and artifacts, was divided into training and testing data sets with a 90% and 10% ratio, respectively. Moreover, in addition to the conventional pixel-wise evaluation metrics, the physics-based metric of the lattice-Boltzmann method (LBM) simulated permeability provided more comprehensive assessments. The results demonstrated that the residual connections, nested architectures, and redesigned skip connections contribute to the model performance and give the residual U-Net++ the highest accuracy. The improvements were mainly on the boundaries and small targets, especially the DFNs, which dominate the interconnectivity and therefore affect the permeability greatly. This study also rigorously evaluated the efficiency and generalization of each model, demonstrating that the sophisticated architectures achieved excellent practicability and maintained robust performance on completely unseen data, ensuring their suitability for diverse and challenging DRP applications.

58 GEOSCIENCES↗