Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Network analysis”

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 937 records · Page 52

Experimental Observations of the Topology of Convolutional Neural Network Activations

Topological data analysis (TDA) is a branch of computational mathematics, bridging algebraic topology and data science, that provides compact, noise-robust representations of complex structures. Deep neural networks (DNNs) learn millions of parameters associated with a series of transformations defined by the model architecture resulting in high-dimensional, difficult to interpret internal representations of input data. As DNNs become more ubiquitous across multiple sectors of our society, there is increasing recognition that mathematical methods are needed to aid analysts, researchers, and practitioners in understanding and interpreting how these models' internal representations relate to the final classification. In this paper we apply cutting edge techniques from TDA with the goal of gaining insight towards interpretability of convolutional neural networks used for image classification. We use two common TDA approaches to explore several methods for modeling hidden layer activations as high-dimensional point clouds, and provide experimental evidence that these point clouds capture valuable structural information about the model's process. First, we demonstrate that a distance metric based on persistent homology can be used to quantify meaningful differences between layers and discuss these distances in the broader context of existing representational similarity metrics for neural network interpretability. Second, we show that a mapper graph can provide semantic insight as to how these models organize hierarchical class knowledge at each layer. These observations demonstrate that TDA is a useful tool to help deep learning practitioners unlock the hidden structures of their models.

topological data analysis, deep learning↗

Industrial PLC Network Modeling and Parameter Identification Using Sensitivity Analysis and Mean Field Variational Inference

A multiple input multiple output (MIMO) power line communication (PLC) model for industrial facilities was developed that uses the physics of a bottom-up model but can be calibrated like top-down models. The PLC model considers 4-conductor cables (three-phase conductors and a ground conductor) and has several load types, including motor loads. The model is calibrated to data using mean field variational inference with a sensitivity analysis to reduce the parameter space. The results show that the inference method can accurately identify many of the model parameters, and the model is accurate even when the network is modified.

47 OTHER INSTRUMENTATION↗

Real-time Intrusion Detection for High-bandwidth Research Networks using Unsupervised Deep Learning

Nowadays, data generated by large-scale scientific experiments are on the scale of petabytes per month. These data are transferred through dedicated high-bandwidth networks across distributed collaboration sites for processing, storage, and analysis. Like general purpose networks, research networks experience attacks. However, monitoring traffic in such high-bandwidth network environments is challenging with conventional deep packet inspection (DPI)- based network intrusion detection systems (NIDS). Recently, machine learning (ML) algorithms have shown potential in detecting anomalous patterns with better generalizability and faster speed. Compared to the general purpose networks, traffic movement in research network is less volatile and more likely to benefit from the ML-based anomaly analysis. We present ThunderSecure, an efficient AI-powered system for detecting anomalies in 100G research networks. ThunderSecure implements a high-throughput packet processing infrastructure using multi-cores and GPUs. It takes time-series statistics from a variety of network measurements and uses 2D Convolutional Neural Network (CNN) to explore both spatial and temporal dependencies in network data stream. Patterns of science data traffic are learned with a one-class neural network, which blends an Adversarial Autoencoder (AAE) with Gaussian mixture density estimation. Testing traffic flows exhibiting significant deviations from the learned baseline of normality are marked as anomalies. We trained ThunderSecure on hundreds of billions of science data packets recorded from a 100G research network at Fermi National Accelerator Laboratory. The detection performance was evaluated on traffic captured from the same research network days and weeks after the training with different types of attack flows injected. To the best of our knowledge, this is the first ML work in the area of network anomaly detection that has been validated on such extreme scale datasets. Results show that ThunderSecure can recognize science data traffic that are captured long after the training and make nearly certain detection on those with anomalous flows injected, even in the case when the anomaly-to-normal mixing ratio is 0.1%.

Gong, Qian↗

Autonomous management of a recursive area hierarchy for large scale wireless sensor networks using multiple parents

Large scale wireless sensor networks have been proposed for applications ranging from anomaly detection in an environment to vehicle tracking. Many of these applications require the networks to be distributed across a large geographic area while supporting three to five year network lifetimes. In order to support these requirements large scale wireless sensor networks of duty-cycled devices need a method of efficient and effective autonomous configuration/maintenance. This method should gracefully handle the synchronization tasks duty-cycled networks. Further, an effective configuration solution needs to recognize that in-network data aggregation and analysis presents significant benefits to wireless sensor network and should configure the network in a way such that said higher level functions benefit from the logically imposed structure. NOA, the proposed configuration and maintenance protocol, provides a multi-parent hierarchical logical structure for the network that reduces the synchronization workload. It also provides higher level functions with significant inherent benefits such as but not limited to: removing network divisions that are created by single-parent hierarchies, guarantees for when data will be compared in the hierarchy, and redundancies for communication as well as in-network data aggregation/analysis/storage.

47 OTHER INSTRUMENTATION↗

Constrained curve fitting for semi-parametric models with radial basis function networks

Common to many analysis pipelines in lattice gauge theory and the broader scientific discipline is the need to fit a semi-parametric model to data. We propose a fit method that utilizes a radial basis function network to approximate the non-parametric component of such models. The approximate parametric model is fit to data using the basin hopping global optimization algorithm. Parameter constraints are enforced through Gaussian priors. The viability of our method is tested by examining its use in a finite-size scaling analysis of the $q$-state Potts model and $p$-state clock model with $q=2,3$ and $p=4,\infty$.

Peterson, Curtis T.↗

A neural network for real-time retrievals of PWV and LWP from Arctic millimeter-wave ground-based observations.

This paper presents a new neural network (NN) algorithm for real-time retrievals of low amounts of precipitable water vapor (PWV) and integrated liquid water from millimeter-wave ground-based observations. Measurements are collected by the 183.3-GHz G-band vapor radiometer (GVR) operating at the Atmospheric Radiation Measurement (ARM) Program Climate Research Facility, Barrow, AK. The NN provides the means to explore the nonlinear regime of the measurements and investigate the physical boundaries of the operability of the instrument. A methodology to compute individual error bars associated with the NN output is developed, and a detailed error analysis of the network output is provided. Through the error analysis, it is possible to isolate several components contributing to the overall retrieval errors and to analyze the dependence of the errors on the inputs. The network outputs and associated errors are then compared with results from a physical retrieval and with the ARM two-channel microwave radiometer (MWR) statistical retrieval. When the NN is trained with a seasonal training data set, the retrievals of water vapor yield results that are comparable to those obtained from a traditional physical retrieval, with a retrieval error percentage of {approx}5% when the PWV is between 2 and 10 mm, but with the advantages that the NN algorithm does not require vertical profiles of temperature and humidity as input and is significantly faster computationally. Liquid water path (LWP) retrievals from the NN have a significantly improved clear-sky bias (mean of {approx}2.4 g/m{sup 2}) and a retrieval error varying from 1 to about 10 g/m{sup 2} when the PWV amount is between 1 and 10 mm. As an independent validation of the LWP retrieval, the longwave downwelling surface flux was computed and compared with observations. The comparison shows a significant improvement with respect to the MWR statistical retrievals, particularly for LWP amounts of less than 60 g/m{sup 2}.

47 OTHER INSTRUMENTATION↗

Structure and thermal behavior of a nickel complex based on a V-shaped bis(4-(1H-imidazol-1-yl)phenyl)amine ligand

A new coordination polymer ([Ni(BIPA)(bpdc)(H{sub 2}O){sub 2}]){sub n} has been prepared based on a new V-shaped bis(4-(1H-imidazol-1-yl)phenyl)amine (BIPA) ligand. Complex was characterized by elemental analysis, IR spectra, X-ray powder diffraction, thermal analysis and single crystal X-ray analysis. Complex exhibits a 1D → 3D structure. Hydrogen bonds play an important role in the formation of supramolecular network structure. Thermal analysis indicates that complex exhibits a high thermal stability.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

The cerebrospinal fluid proteome in HIV infection: change associated with disease severity.

Central nervous system (CNS) infection is a constant feature of systemic HIV infection with a clinical spectrum that ranges from chronic asymptomatic infection to severe cognitive and motor dysfunction. Analysis of cerebrospinal fluid (CSF) has played an important part in defining the character of this evolving infection and response to treatment. To further characterize CNS HIV infection and its effects, we applied advanced high-throughput proteomic methods to CSF to identify novel proteins and their changes with disease progression and treatment. After establishing an accurate mass and time (AMT) tag database containing 23,141 AMT tags for CSF peptides, we analyzed 91 CSF samples by LC-MS from 12 HIV-uninfected and 14 HIV-infected subjects studied in the context of initiation of antiretroviral and correlated abundances of identified proteins (a) within and between subjects, (b) with all other proteins across the entire sample set, and (c) with 'external' CSF biomarkers of infection (HIV RNA), immune activation (neopterin) and neural injury (neurofilament light chain protein, NFL). We identified a mean of 2,333 +/- 328 (SD) peptides covering 307 +/-16 proteins in the 91 CSF sample set. Protein abundances differed both between and within subjects sampled at different time points and readily separated those with and without HIV infection. Proteins also showed inter-correlations across the sample set that were associated with biologically relevant dynamic processes. One-hundred and fifty proteins showed correlations with the external biomarkers. For example, using a threshold of cross correlation coefficient (Pearson's) {le}0.3 and {ge}0.3 for potentially meaningful relationships, a total of 99 proteins correlated with CSF neopterin (43 negative and 56 positive correlations) and related principally to neuronal plasticity and survival and to innate immunity. Pathway analysis defined several networks connecting the identified proteins, including one with amyloid precursor protein as a central node. Advanced CSF proteomic analysis enabled the identification of an array of novel protein changes across the spectrum of CNS HIV infection and disease. This initial analysis clearly demonstrated the value of contemporary state-of-the-art proteomic CSF analysis as a discovery tool in HIV infection with likely similar application to other neurological inflammatory and degenerative diseases.

59 BASIC BIOLOGICAL SCIENCES↗

REMODEL: Rethinking Deep CNN Models to Detect and Count on a NeuroSynaptic System

In this work, we perform analysis of detection and counting of cars using a low-power IBM TrueNorth Neurosynaptic System. For our evaluation we looked at a publicly-available dataset that has overhead imagery of cars with context present in the image. The trained neural network for image analysis was deployed on the NS16e system using IBM's EEDN training framework. Through multiple experiments we identify the architectural bottlenecks present in TrueNorth system that does not let us deploy large neural network structures. Following these experiments we propose changes to CNN model to circumvent these architectural bottlenecks. The results of these evaluations have been compared with caffe-based implementations of standard neural networks that were deployed on a Titan-X GPU. Results showed that TrueNorth can detect cars from the dataset with 97.60% accuracy and can be used to accurately count the number of cars in the image with 69.04% accuracy. The car detection accuracy and car count (–/+ 2 error margin) accuracy are comparable to high-precision neural networks like AlexNet, GoogLeNet, and ResCeption, but show a manifold improvement in power consumption.

97 MATHEMATICS AND COMPUTING↗

Forced synchronization of autonomous dynamical Boolean networks

We present the design of an autonomous time-delay Boolean network realized with readily available electronic components. Through simulations and experiments that account for the detailed nonlinear response of each circuit element, we demonstrate that a network with five Boolean nodes displays complex behavior. Furthermore, we show that the dynamics of two identical networks display near-instantaneous synchronization to a periodic state when forced by a common periodic Boolean signal. A theoretical analysis of the network reveals the conditions under which complex behavior is expected in an individual network and the occurrence of synchronization in the forced networks. This research will enable future experiments on autonomous time-delay networks using readily available electronic components with dynamics on a slow enough time-scale so that inexpensive data collection systems can faithfully record the dynamics.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DECODING SPECTRAL ENERGY DISTRIBUTIONS OF DUST-OBSCURED STARBURST-ACTIVE GALACTIC NUCLEUS

We present BayeSED, a general purpose tool for Bayesian analysis of spectral energy distributions (SEDs) using pre-existing model SED libraries or their linear combinations. The artificial neural networks, principal component analysis, and multimodal-nested sampling (MultiNest) techniques are employed to allow the highly efficient sampling of posterior distribution and the calculation of Bayesian evidence. As a demonstration, we apply this tool to a sample of hyperluminous infrared galaxies (HLIRGs). The Bayesian evidence obtained for a pure starburst, a pure active galactic nucleus (AGN), and a linear combination of starburst+AGN models show that the starburst+AGN model has the highest evidence for all galaxies in this sample. The Bayesian evidence for the three models and the estimated contributions of starbursts and AGNs to infrared luminosity show that HLIRGs can be classified into two groups: one dominated by starbursts and the other dominated by AGNs. Other parameters and corresponding uncertainties about starbursts and AGNs are also estimated using the model with the highest Bayesian evidence. We find that the starburst region of the HLIRGs dominated by starbursts tends to be more compact and has a higher fraction of OB stars than that of HLIRGs dominated by AGNs. Meanwhile, the AGN torus of the HLIRGs dominated by AGNs tends to be more dusty than that of HLIRGs dominated by starbursts. These results are consistent with previous researches, but need to be tested further with larger samples. Overall, we believe that BayeSED could be a reliable and efficient tool for exploring the nature of complex systems such as dust-obscured starburst-AGN composite systems by decoding their SEDs.

79 ASTRONOMY AND ASTROPHYSICS↗

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)↗

Stability-driven nonnegative matrix factorization to interpret spatial gene expression and build local gene networks

Spatial gene expression patterns enable the detection of local covariability and are extremely useful for identifying local gene interactions during normal development. The abundance of spatial expression data in recent years has led to the modeling and analysis of regulatory networks. The inherent complexity of such data makes it a challenge to extract biological information. We developed staNMF, a method that combines a scalable implementation of nonnegative matrix factorization (NMF) with a new stability-driven model selection criterion. When applied to a set of Drosophila early embryonic spatial gene expression images, one of the largest datasets of its kind, staNMF identified 21 principal patterns (PP). Providing a compact yet biologically interpretable representation of Drosophila expression patterns, PP are comparable to a fate map generated experimentally by laser ablation and show exceptional promise as a data-driven alternative to manual annotations. Our analysis mapped genes to cell-fate programs and assigned putative biological roles to uncharacterized genes. Finally, we used the PP to generate local transcription factor regulatory networks. Spatially local correlation networks were constructed for six PP that span along the embryonic anterior-posterior axis. Using a two-tail 5% cutoff on correlation, we reproduced 10 of the 11 links in the well-studied gap gene network. In conclusion, the performance of PP with the Drosophila data suggests that staNMF provides informative decompositions and constitutes a useful computational lens through which to extract biological insight from complex and often noisy gene expression data.

59 BASIC BIOLOGICAL SCIENCES↗

Evaluating conducting network based transparent electrodes from geometrical considerations

Conducting nanowire networks have been developed as viable alternative to existing indium tin oxide based transparent electrode (TE). The nature of electrical conduction and process optimization for electrodes have gained much from the theoretical models based on percolation transport using Monte Carlo approach and applying Kirchhoff's law on individual junctions and loops. While most of the literature work pertaining to theoretical analysis is focussed on networks obtained from conducting rods (mostly considering only junction resistance), hardly any attention has been paid to those made using template based methods, wherein the structure of network is neither similar to network obtained from conducting rods nor similar to well periodic geometry. Here, we have attempted an analytical treatment based on geometrical arguments and applied image analysis on practical networks to gain deeper insight into conducting networked structure particularly in relation to sheet resistance and transmittance. Many literature examples reporting networks with straight or curvilinear wires with distributions in wire width and length have been analysed by treating the networks as two dimensional graphs and evaluating the sheet resistance based on wire density and wire width. The sheet resistance values from our analysis compare well with the experimental values. Our analysis on various examples has revealed that low sheet resistance is achieved with high wire density and compactness with straight rather than curvilinear wires and with narrower wire width distribution. Similarly, higher transmittance for given sheet resistance is possible with narrower wire width but of higher thickness, minimal curvilinearity, and maximum connectivity. For the purpose of evaluating active fraction of the network, the algorithm was made to distinguish and quantify current carrying backbone regions as against regions containing only dangling or isolated wires. The treatment can be helpful in predicting the properties of a network simply from image analysis and will be helpful in improvisation and comparison of various TEs and better understanding of electrical percolation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

G-NetMon: a GPU-accelerated network performance monitoring system

At Fermilab, we have prototyped a GPU-accelerated network performance monitoring system, called G-NetMon, to support large-scale scientific collaborations. In this work, we explore new opportunities in network traffic monitoring and analysis with GPUs. Our system exploits the data parallelism that exists within network flow data to provide fast analysis of bulk data movement between Fermilab and collaboration sites. Experiments demonstrate that our G-NetMon can rapidly detect sub-optimal bulk data movements.

99 GENERAL AND MISCELLANEOUS↗

Global sensitivity analysis in stochastic simulators of uncertain reaction networks

Stochastic models of chemical systems are often subjected to uncertainties in kinetic parameters in addition to the inherent random nature of their dynamics. Uncertainty quantification in such systems is generally achieved by means of sensitivity analyses in which one characterizes the variability with the uncertain kinetic parameters of the first statistical moments of model predictions. In this work, we propose an original global sensitivity analysis method where the parametric and inherent variability sources are both treated through Sobol’s decomposition of the variance into contributions from arbitrary subset of uncertain parameters and stochastic reaction channels. The conceptual development only assumes that the inherent and parametric sources are independent, and considers the Poisson processes in the random-time-change representation of the state dynamics as the fundamental objects governing the inherent stochasticity. Here, a sampling algorithm is proposed to perform the global sensitivity analysis, and to estimate the partial variances and sensitivity indices characterizing the importance of the various sources of variability and their interactions. The birth-death and Schlögl models are used to illustrate both the implementation of the algorithm and the richness of the proposed analysis method. The output of the proposed sensitivity analysis is also contrasted with a local derivative-based sensitivity analysis method classically used for this type of systems.

97 MATHEMATICS AND COMPUTING↗

Real-Time SCADA Cyber Protection Using Compression Techniques

The Department of Energy’s Office of Electricity Delivery and Energy Reliability (DOE-OE) has a critical mission to secure the energy infrastructure from cyber attack. Through DOE-OE’s Cybersecurity for Energy Delivery Systems (CEDS) program, the Idaho National Laboratory (INL) has developed a method to detect malicious traffic on Supervisory, Control, and Data Acquisition (SCADA) network using a data compression technique. SCADA network traffic is often repetitive with only minor differences between packets. Research performed at the INL showed that SCADA network traffic has traits desirable for using compression analysis to identify abnormal network traffic. An open source implementation of a Lempel-Ziv-Welch (LZW) lossless data compression algorithm was used to compress and analyze surrogate SCADA traffic. Infected SCADA traffic was found to have statistically significant differences in compression when compared against normal SCADA traffic at the packet level. The initial analyses and results are clearly able to identify malicious network traffic from normal traffic at the packet level with a very high confidence level across multiple ports and traffic streams. Statistical differentiation between infected and normal traffic level was possible using a modified data compression technique at the 99% probability level for all data analyzed. However, the conditions tested were rather limited in scope and need to be expanded into more realistic simulations of hacking events using techniques and approaches that are better representative of a real-world attack on a SCADA system. Nonetheless, the use of compression techniques to identify malicious traffic on SCADA networks in real time appears to have significant merit for infrastructure protection.

97 MATHEMATICS AND COMPUTING↗

Statistical data analysis of x-ray spectroscopy data enabled by neural network accelerated Bayesian inference

Bayesian inference applied to x-ray spectroscopy data analysis enables uncertainty quantification necessary to rigorously test theoretical models. However, when comparing to data, detailed atomic physics and radiation transfer calculations of x-ray emission from non-uniform plasma conditions are typically too slow to be performed in line with statistical sampling methods, such as Markov Chain Monte Carlo sampling. Furthermore, differences in transition energies and x-ray opacities often make direct comparisons between simulated and measured spectra unreliable. Here, we present a spectral decomposition method that allows for corrections to line positions and bound–bound opacities to best fit experimental data, with the goal of providing quantitative feedback to improve the underlying theoretical models and guide future experiments. In this work, we use a neural network (NN) surrogate model to replace spectral calculations of isobaric hot-spots created in Kr-doped implosions at the National Ignition Facility. The NN was trained on calculations of x-ray spectra using an isobaric hot-spot model post-processed with Cretin, a multi-species atomic kinetics and radiation code. The speedup provided by the NN model to generate x-ray emission spectra enables statistical analysis of parameterized models with sufficient detail to accurately represent the physical system and extract the plasma parameters of interest.

47 OTHER INSTRUMENTATION↗