Engineering Papers⌕ Search

SEARCH · Engineering Papers

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 595 records · Page 33
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Multiscale Flow for robust and optimal cosmological analysis

We propose Multiscale Flow, a generative Normalizing Flow that creates samples and models the field-level likelihood of two-dimensional cosmological data such as weak lensing. Multiscale Flow uses hierarchical decomposition of cosmological fields via a wavelet basis and then models different wavelet components separately as Normalizing Flows. The log-likelihood of the original cosmological field can be recovered by summing over the log-likelihood of each wavelet term. This decomposition allows us to separate the information from different scales and identify distribution shifts in the data such as unknown scale-dependent systematics. The resulting likelihood analysis can not only identify these types of systematics, but can also be made optimal, in the sense that the Multiscale Flow can learn the full likelihood at the field without any dimensionality reduction. We apply Multiscale Flow to weak lensing mock datasets for cosmological inference and show that it significantly outperforms traditional summary statistics such as power spectrum and peak counts, as well as machine learning–based summary statistics such as scattering transform and convolutional neural networks. We further show that Multiscale Flow is able to identify distribution shifts not in the training data such as baryonic effects. Finally, we demonstrate that Multiscale Flow can be used to generate realistic samples of weak lensing data.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Electrochemical Random-Access Memory: Progress, Perspectives, and Opportunities

Non-von Neumann computing using neuromorphic systems based on analogue synaptic and neuronal elements has emerged as a potential solution to tackle the growing need for more efficient data processing, but progress toward practical systems has been stymied due to a lack of materials and devices with the appropriate attributes. Recently, solid state electrochemical ion-insertion, also known as electrochemical random access memory (ECRAM) has emerged as a promising approach to realize the needed device characteristics. ECRAM is a three terminal device that operates by tuning electronic conductance in functional materials through solid-state electrochemical redox reactions. This mechanism can be considered as a gate-controlled bulk modulation of dopants and/or phases in the channel. Early work demonstrating that ECRAM can achieve nearly ideal analogue synaptic characteristics has sparked tremendous interest in this approach. More recently, the realization that electrochemical ion insertion can be used to tune the electronic properties of many types of materials including transition metal oxides, layered two-dimensional materials, organic and coordination polymers, and that the changes in conductance can span orders of magnitude has further attracted interest in ECRAM as the basis for analogue synaptic elements for inference accelerators as well as for dynamical devices that can emulate a wide range of neuronal characteristics for implementation in analogue spiking neural networks. At its core, ECRAM shares many fundamental aspects with rechargeable batteries, where ion insertion materials are used extensively for their ability to reversibly store charge and energy. Computing applications, however, present drastically different requirements: systems will require many millions of devices, scaled down to tens of nanometers, all while achieving reliable electronic-state tuning at scaled-up rates and endurances, and with minimal energy dissipation and noise. Further, in this review, we discuss the history, basic concepts, recent progress, as well as the challenges and opportunities for different types of ECRAM, broadly grouped by their primary mobile ionic charge carrier, including Li, protons, and oxygen vacancies.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Computational design of highly signalling-active membrane receptors through solvent-mediated allosteric networks

Abstract Protein catalysis and allostery require the atomic-level orchestration and motion of residues and ligand, solvent and protein effector molecules. However, the ability to design protein activity through precise protein–solvent cooperative interactions has not yet been demonstrated. Here we report the design of 14 membrane receptors that catalyse G protein nucleotide exchange through diverse engineered allosteric pathways mediated by cooperative networks of intraprotein, protein–ligand and –solvent molecule interactions. Consistent with predictions, the designed protein activities correlated well with the level of plasticity of the networks at flexible transmembrane helical interfaces. Several designs displayed considerably enhanced thermostability and activity compared with related natural receptors. The most stable and active variant crystallized in an unforeseen signalling-active conformation, in excellent agreement with the design models. The allosteric network topologies of the best designs bear limited similarity to those of natural receptors and reveal an allosteric interaction space larger than previously inferred from natural proteins. The approach should prove useful for engineering proteins with novel complex protein binding, catalytic and signalling activities.

Chemistry↗
DOE OSTI2025

Neural network-based model of galaxy power spectrum: fast full-shape galaxy power spectrum analysis

ABSTRACT We present a neural network-based emulator for the galaxy redshift-space power spectrum that enables several orders of magnitude acceleration in the galaxy clustering parameter inference, while preserving 3$\sigma$ accuracy better than 0.5 per cent up to $k_{\mathrm{max}}$ = 0.25 $\, h\text{Mpc}^{-1}$ within Lambda-cold dark matter ($\Lambda$CDM) and around 0.5 per cent $w_0$–$w_a$CDM. Our surrogate model only emulates the galaxy bias-invariant terms of one-loop perturbation theory predictions, these terms are then combined analytically with galaxy bias terms, counter-terms, and stochastic terms in order to obtain the non-linear redshift-space galaxy power spectrum. This allows us to avoid any galaxy bias prescription in the training of the emulator, which makes it more flexible. Moreover, we include the redshift $z \in [0,1.4]$ in the training which further avoids the need for re-training the emulator. We showcase the performance of the emulator in recovering the cosmological parameters of $\Lambda$CDM by analysing the suite of 25 AbacusSummit simulations that mimic the Dark Energy Spectroscopic Instrument luminous red galaxies at $z=0.5$ and 0.8, together as the emission line galaxies at $z=0.8$. We obtain similar performance in all cases, demonstrating the reliability of the emulator for any galaxy sample at any redshift in $0 \lt z \lt 1.4$. We will make our emulator public at github repository.

Trusov, Svyatoslav (ORCID:0000000224146720)↗
DOE OSTI2023

Semi-supervised graph neural networks for pileup noise removal

Abstract The high instantaneous luminosity of the CERN Large Hadron Collider leads to multiple proton–proton interactions in the same or nearby bunch crossings (pileup). Advanced pileup mitigation algorithms are designed to remove this noise from pileup particles and improve the performance of crucial physics observables. This study implements a semi-supervised graph neural network for particle-level pileup noise removal, by identifying individual particles produced from pileup. The graph neural network is firstly trained on charged particles with known labels, which can be obtained from detector measurements on data or simulation, and then inferred on neutral particles for which such labels are missing. This semi-supervised approach does not depend on the neutral particle pileup label information from simulation, and thus allows us to perform training directly on experimental data. The performance of this approach is found to be consistently better than widely-used domain algorithms and comparable to the fully-supervised training using simulation truth information. The study serves as the first attempt at applying semi-supervised learning techniques to pileup mitigation, and opens up a new direction of fully data-driven machine learning pileup mitigation studies.

43 PARTICLE ACCELERATORS↗
DOE OSTI2024

Sentinel

Network intrusion detection systems (NIDS) are commonplace in network security but they frequently employ algorithms that are computational demanding requiring hardware and software with significant power requirements. Two examples of such resource-intensive algorithms used for network security are regular expression matching and broader signature pattern matching which are commonly used in deep packet inspection (DPI). Network security algorithms that have large power requirements may be a challenge for low-power internet-of-things (IoT) environments, which generally lack the power resources to implement complex security measures like computationally expensive DPI at the edge. Furthermore, IoT environments incorporating 5G standalone networks have network latency constraints beyond just power that make DPI at the edge even more difficult. Programmable logic is ideally suited for machine learning inference for DPI because of its deep instruction level parallelism and single-cycle memory access. Machine learning approaches for DPI have been explored before using the programmable logic of field programmable gate arrays (FPGA) as a potential solution for NIDS approaches that would be power-suitable for IoT. However, those previous programmable logic NIDS approaches utilize either a supervised or unsupervised learning model. Sentinel utilizes the ensemble of these two machine learning approaches known as a semi-supervised approach which has shown promise in NIDS implementations. Sentinel provides a programmable logic implementation of a semi-supervised approach for DPI which operates at much lower power and latency than a GPU implementation with negligible loss of accuracy due to quantization through a logistic regressor.

Anderson, MatthewW [Idaho National Laboratory (INL↗
NASA NTRSDate not supplied

Advances in Entry Modeling for Impact Risk Assessment

A summary of recent advancements in the detailed modeling of asteroid atmospheric entry processes made through NASA’s Asteroid Threat Assessment Project (ATAP) is presented. Understanding, and accurately modeling these processes and their associated uncertainties is critical to predicting all downstream impact effects, such as blast wave and thermal damage footprints. Furthermore, there is (perhaps thankfully) a dearth of empirical data for large impactors of the kind that would pose a threat to human populations, on which to anchor and/or validate models used in risk assessments. Therefore, we must rely heavily on detailed theoretical and numerical modeling to develop robust assessments for decision makers. To that end, ATAP has made some significant progress in advancing the capabilities in this area. Two areas in particular are highlighted in the present work: meteoroid ablation mechanisms, and bolide luminosity. The first research area – meteoroid ablation mechanisms – has focused on performing novel high-enthalpy wind tunnel experiments on meteorites and meteorite analogs, and utilizing the resulting data to develop high-fidelity models for impactor mass loss at scale. These efforts have resulted in several insights. Of note, these data suggest a differential vaporization process where volatiles are liberated preferentially when the asteroidal material is subject to high heat, while refractory components remain in the molten layer on the surface. A numerical model has been developed which considers this phenomena, and its effect on the bulk impactors effective heat of ablation is examined. The second research thrust that is discussed is focused on accurate modeling of impactor radiation phenomena. While another submission to the conference will discuss the application of our approach to thermal ground damage modeling, here, we present an overview of our extensive efforts to utilize available ground- and space-based observations of large bolides (~1m diameter, and above) to inform and validate our detailed modeling approaches. These methods have been shown to accurately reconstruct the detailed spectra for the Benesov bolide, as well as approximate the burn footprint for the Tunguska event. Recent effort has focused on reconciling light curve data from multiple sources (all-sky camera networks, GLM, US government sensors), using our validated model, and providing a calibrated model for luminous efficiency which can then be utilized to infer impactor properties such as shape, mass, and composition. This work will be demonstrated through an exemplar case study focusing on a large bolide event with observational data (e.g. Chelyabinsk, Flensburg, etc.). Finally, our team’s assessment on the current maturity of atmospheric entry models, and priorities for future research will be provided.

ATAP↗
DOE OSTI2024

Graph Neural Network-based Tracking as a Service

Recent studies have shown promising results for track finding in dense environments using Graph Neural Network (GNN)-based algorithms. However, GNN-based track finding is computationally slow on CPUs, necessitating the use of coprocessors to accelerate the inference time. Additionally, the large input graph size demands a large device memory for efficient computation, a requirement not met by all computing facilities used for particle physics experiments, particularly those lacking advanced GPUs. Furthermore, deploying the GNN-based track-finding algorithm in a production environment requires the installation of all dependent software packages, exclusively utilized by this algorithm. These computing challenges must be addressed for the successful implementation of GNN-based track-finding algorithm into production settings. In response, we introduce a ``GNN-based tracking as a service'' approach, incorporating a custom backend within the NVIDIA Triton inference server to facilitate GNN-based tracking. This paper presents the performance of this approach using the Perlmutter supercomputer at NERSC.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Even Higher-Level Synthesis: An Exploration of AI Hardware Accelerators using HLS4ML

With the rise of artificial intelligence, the popularization of deep learning, and a constantly evolving industry, the demand for flexible and efficient tools has never been greater. As algorithms grow more complex, their runtime and energy consumption increase exponentially. Customized hardware accelerators, long used for specific mathematical operations, remain essential for managing modern applications' computational and power demands. Hardware accelerators can speed up complex computations by orders of magnitude, but their manual design and verification processes are often challenging and time-consuming. High-Level Synthesis (HLS) provides a solution by transforming high-level algorithm descriptions, typically written in C++ or SystemC, into synthesizable RTL suitable for hardware implementation. This approach reduces development time for RTL engineers while offering flexibility beyond what traditional handwritten RTL can provide. We extended this capability to the machine-learning domain with the open-source framework hls4ml, which allows neural networks trained in Python frameworks like Tensorflow or PyTorch to be synthesized into efficient hardware representations for the traditional FPGA and ASIC flows. This breakthrough addresses the growing need for reduced design turnaround and easy verification of ML hardware accelerators with low latency and power efficiency constraints. During this tutorial, we will demonstrate how Python complements HLS by simplifying the ML design process, bridging the gap between software and hardware development. Attendees will explore how we translate neural networks modeled in Python into fixed-point C++ models suitable for HLS workflows. We will dive into strategies like Value-Range Analysis and Quantization-Aware Training, which optimize these designs for deployment and evaluate their accuracy, power consumption, and energy efficiency. To exemplify these concepts, experts from Fermilab will share their experiences applying this technology to high-energy physics experiments, where real-time, low-latency processing is critical. Over the years, Fermilab engineers have demonstrated how deep neural networks, optimized for hardware using hls4ml, can meet the stringent requirements of trigger systems at the CERN Large Hadron Collider. These systems rely on rapid decision-making to process immense data volumes while retaining only the most relevant events for further analysis. The application of hls4ml has also been extended to innovative technologies like smart pixel arrays. These smart pixels integrate ML inference capabilities directly into sensor devices, enabling localized data processing at the pixel level. This approach drastically reduces the need to transmit raw data to external processing units, significantly decreasing power consumption and latency. By embedding neural networks within the pixel architecture, the smart pixels can identify and prioritize relevant data in real time, providing a highly efficient solution for edge computing in scenarios such as particle detectors and imaging systems. Fermilab's work highlights the potential of hardware-accelerated ML in scenarios where both speed and power efficiency are mission-critical. Through this tutorial, attendees will gain valuable insights into the challenges and solutions of deploying ML in hardware. Understanding how HLS and hls4ml streamline the development of neural network-based hardware accelerators is fundamental for the industry's future. Participants will learn how these technologies are shaping the future of AI and scientific computing.

Di Guglielmo, Giuseppe [Fermilab]↗
NASA NTRS2000

Filtering in Hybrid Dynamic Bayesian Networks

We implement a 2-time slice dynamic Bayesian network (2T-DBN) framework and make a 1-D state estimation simulation, an extension of the experiment in (v.d. Merwe et al., 2000) and compare different filtering techniques. Furthermore, we demonstrate experimentally that inference in a complex hybrid DBN is possible by simulating fault detection in a watertank system, an extension of the experiment in (Koller & Lerner, 2000) using a hybrid 2T-DBN. In both experiments, we perform approximate inference using standard filtering techniques, Monte Carlo methods and combinations of these. In the watertank simulation, we also demonstrate the use of 'non-strict' Rao-Blackwellisation. We show that the unscented Kalman filter (UKF) and UKF in a particle filtering framework outperform the generic particle filter, the extended Kalman filter (EKF) and EKF in a particle filtering framework with respect to accuracy in terms of estimation RMSE and sensitivity with respect to choice of network structure. Especially we demonstrate the superiority of UKF in a PF framework when our beliefs of how data was generated are wrong. Furthermore, we investigate the influence of data noise in the watertank simulation using UKF and PFUKD and show that the algorithms are more sensitive to changes in the measurement noise level that the process noise level. Theory and implementation is based on (v.d. Merwe et al., 2000).

Andersen, Morten Nonboe↗
DOE OSTI2024

Inferring the Focal Depths of Small Earthquakes in Southern California Using Physics-Based Waveform Features

Determining the depths of small crustal earthquakes is challenging in many regions of the world, because most seismic networks are too sparse to resolve trade-offs between depth and origin time with conventional arrival-time methods. Precise and accurate depth estimation is important, because it can help seismologists discriminate between earthquakes and explosions, which is relevant to monitoring nuclear test ban treaties and producing earthquake catalogs that are uncontaminated by mining blasts. Here, we examine the depth sensitivity of several physics-based waveform features for ~8000 earthquakes in southern California that have well-resolved depths from arrival-time inversion. We focus on small earthquakes (2 < M L < 4) recorded at local distances (<150 km), for which depth estimation is especially challenging. We find that differential magnitudes (M w /M L –M c ) are positively correlated with focal depth, implying that coda wave excitation decreases with focal depth. We analyze a simple proxy for relative frequency content, Φ≡log 10 (M 0 )+3log 10 (f c ), and find that source spectra are preferentially enriched in high frequencies, or “blue-shifted,” as focal depth increases. Here, we also find that two spectral amplitude ratios Rg 0.5–2 Hz/Sg 0.5–8 Hz and Pg/Sg at 3–8 Hz decrease as focal depth increases. Using multilinear regression with these features as predictor variables, we develop models that can explain 11%–59% of the variance in depths within 10 subregions and 25% of the depth variance across southern California as a whole. We suggest that incorporating these features into a machine learning workflow could help resolve focal depths in regions that are poorly instrumented and lack large databases of well-located events. Some of the waveform features we evaluate in this study have previously been used as source discriminants, and our results imply that their effectiveness in discrimination is partially because explosions generally occur at shallower depths than earthquakes.

58 GEOSCIENCES↗
DOE OSTI2022

Open-source FPGA-ML codesign for the MLPerf Tiny Benchmark

We present our development experience and recent results for the MLPerf Tiny Inference Benchmark on field-programmable gate array (FPGA) platforms. We use the open-source hls4ml and FINN workflows, which aim to democratize AI-hardware codesign of optimized neural networks on FPGAs. We present the design and implementation process for the keyword spotting, anomaly detection, and image classification benchmark tasks. The resulting hardware implementations are quantized, configurable, spatial dataflow architectures tailored for speed and efficiency and introduce new generic optimizations and common workflows developed as a part of this work. The full workflow is presented from quantization-aware training to FPGA implementation. The solutions are deployed on system-on-chip (Pynq-Z2) and pure FPGA (Arty A7-100T) platforms. The resulting submissions achieve latencies as low as 20 $\mu$s and energy consumption as low as 30 $\mu$J per inference. We demonstrate how emerging ML benchmarks on heterogeneous hardware platforms can catalyze collaboration and the development of new techniques and more accessible tools.

Borras, Hendrik↗
DOE OSTI2023

Fast uncertainty estimates in deep learning interatomic potentials

Deep learning has emerged as a promising paradigm to give access to highly accurate predictions of molecular and material properties. A common short-coming shared by current approaches, however, is that neural networks only give point estimates of their predictions and do not come with predictive uncertainties associated with these estimates. Existing uncertainty quantification efforts have primarily leveraged the standard deviation of predictions across an ensemble of independently trained neural networks. This incurs a large computational overhead in both training and prediction, resulting in order-of-magnitude more expensive predictions. Here, we propose a method to estimate the predictive uncertainty based on a single neural network without the need for an ensemble. This allows us to obtain uncertainty estimates with virtually no additional computational overhead over standard training and inference. We demonstrate that the quality of the uncertainty estimates matches those obtained from deep ensembles. We further examine the uncertainty estimates of our methods and deep ensembles across the configuration space of our test system and compare the uncertainties to the potential energy surface. Finally, we study the efficacy of the method in an active learning setting and find the results to match an ensemble-based strategy at order-of-magnitude reduced computational cost.

Chemistry↗
DOE OSTI2020

Framework For Spatial Agricultural Crop Yield Prediction Model Development

This framework was developed to provide data preprocessing for spatiotemporal agricultural yield data and remote sensing data for modelling using artificial neural networks (ANNs) to predict subfield crop yield estimates. The software includes methods to train, validate, and test ANN models. It also include methods to infer on new remote sensing data.

Griffel, LloydM.↗
DOE OSTI2025

FPGA-accelerated SpeckleNN with SNL for real-time X-ray single-particle imaging

We present the implementation of a specialized version of our previously published unified embedding model, SpeckleNN, for real-time speckle pattern classification in X-ray Single-Particle Imaging (SPI), using the SLAC Neural Network Library (SNL) on an FPGA platform. This hardware realization transitions SpeckleNN from a prototypic model into a practical edge solution, optimized for running inference near the detector in high-throughput X-ray free-electron laser (XFEL) facilities, such as those found at the Linac Coherent Light Source (LCLS). To address the resource constraints inherent in FPGAs, we developed a more specialized version of SpeckleNN. The original model, which was designed for broader classification across multiple biological samples, comprised ~5.6 million parameters. The new implementation, while reducing the parameter count to 64.6K (a 98.8% reduction), focuses on maintaining the model's essential functionality for real-time operation, achieving an accuracy of 90%. Furthermore, we compressed the latent space from 128 to 50 dimensions. This implementation was demonstrated on the KCU1500 FPGA board, utilizing 71% of available DSPs, 75% of LUTs, and 48% of FFs, with an average power consumption of 9.4W according to the Vivado post-implementation report. The FPGA performed inference on a single image with a latency of 45.015 microseconds at a 200 MHz clock rate. In comparison, running the same inference on an NVIDIA A100 GPU resulted in an average power consumption of ~73W and an image processing latency of around 400 microseconds. Our FPGA-accelerated version of SpeckleNN demonstrated significant improvements, achieving an 8.9 × speedup and a 7.8 × reduction in power consumption compared to the GPU implementation. Key advancements include model specialization and dynamic weight loading through SNL, which eliminates the need for time-consuming FPGA design re-synthesis, allowing fast and continuous deployment of models (re)trained online. These innovations enable real-time adaptive classification and efficient vetoing of speckle patterns, making SpeckleNN more suited for deployment in XFEL facilities. This implementation has the potential to significantly accelerate SPI experiments and enhance adaptability to evolving experimental conditions.

47 OTHER INSTRUMENTATION↗
DOE OSTI2025

Joint Modeling of Quasar Variability and Accretion Disk Reprocessing Using Latent Stochastic Differential Equations

Quasars are bright active galactic nuclei powered by the accretion of matter around supermassive black holes at the center of galaxies. Their stochastic brightness variability depends on the physical properties of the accretion disk and black hole. The upcoming Rubin Observatory Legacy Survey of Space and Time (LSST) is expected to observe tens of millions of quasars, so there is a need for efficient techniques like machine learning that can handle the large volume of data. Quasar variability is believed to be driven by an X-ray corona, which is reprocessed by the accretion disk and emitted as UV/optical variability. We are the first to introduce an auto-differentiable simulation of the accretion disk and reprocessing. We use the simulation as a direct component of our neural network to jointly model the driving variability and reprocessing, trained with supervised learning on simulated LSST-like 10 yr quasar light curves. We encode the light curves using a transformer encoder, and the driving variability is reconstructed using latent stochastic differential equations, a physically motivated generative deep learning method that can model continuous-time stochastic dynamics. By embedding the physical processes of the driving signal and reprocessing into our network, we achieve a model that is more robust and interpretable. We demonstrate that our model outperforms a Gaussian process regression baseline and can infer accretion disk parameters and time delays between wave bands, even for out-of-distribution driving signals. Our approach provides a powerful framework that can be adapted to solve other inverse problems in multivariate time series.

Fagin, Joshua [City Univ. of New York (CUNY), NY (↗
NASA NTRS2012

Estimation and Bias Correction of Aerosol Abundance using Data-driven Machine Learning and Remote Sensing

Air quality information is increasingly becoming a public health concern, since some of the aerosol particles pose harmful effects to peoples health. One widely available metric of aerosol abundance is the aerosol optical depth (AOD). The AOD is the integrated light extinction coefficient over a vertical atmospheric column of unit cross section, which represents the extent to which the aerosols in that vertical profile prevent the transmission of light by absorption or scattering. The comparison between the AOD measured from the ground-based Aerosol Robotic Network (AERONET) system and the satellite MODIS instruments at 550 nm shows that there is a bias between the two data products. We performed a comprehensive analysis exploring possible factors which may be contributing to the inter-instrumental bias between MODIS and AERONET. The analysis used several measured variables, including the MODIS AOD, as input in order to train a neural network in regression mode to predict the AERONET AOD values. This not only allowed us to obtain an estimate, but also allowed us to infer the optimal sets of variables that played an important role in the prediction. In addition, we applied machine learning to infer the global abundance of ground level PM2.5 from the AOD data and other ancillary satellite and meteorology products. This research is part of our goal to provide air quality information, which can also be useful for global epidemiology studies.

Malakar, Nabin K.↗
DOE OSTI2020

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↗