Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Selective classification”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,009 records · Page 56

HAIMOS Ensemble Forecasts for Intra-day and Day- Ahead GHI, DNI and Ramps

The objective of this research is to develop a hybrid physics-based/data-driven forecast model to improve direct normal and global horizontal irradiance (DNI and GHI) prediction for horizons ranging from 1 to 72 hours. Project objectives also address key gaps in state-of-the-art solar forecasting: accurate probabilistic solar forecasts and the forecasting of large irradiance ramps (ramp onset and magnitude). The proposed model ensembles Numerical Weather Prediction (NWP) forecasts, determinist physics-based algorithms, and new-generation cloud cover products (high-resolution rapid refresh satellite images and Large Eddy Simulations). The result is the Hybrid Adaptive Input Model Objective Selection (HAIMOS) ensemble model. HAIMOS blends state of the art machine learning methodologies with physics-based models for cloud cover and cloud optical depth forecasts. The technical activities followed a two-pronged strategy. First, the preprocessing of data, the selection of inputs to the nonlinear approximators, the type of approximator and objective functions, and post-processing ensembling techniques included in HAIMOS were all optimized adaptively to find the best model for a specific goal (reduce DNI/GHI forecast error, improve the prediction of ramp onset, etc.). Second, a large effort was put in improving cloud identification and the forecast of cloud cover and cloud optical depth. To this end, new-generation cloud parametrization products were developed in this work. These include improved algorithms to assist in cloud identification, cloud classification and cloud parametrization from satellite images – three key factors in the accuracy of 1 to 6-hours irradiance forecasts and prediction of ramp onset. Furthermore, we also included cloud information extracted from high resolution rapid refresh satellite images (GOES-16) and Large Eddy Simulations (LES). LES was used to model the atmosphere in detail over locations of interest and produce cloud optical depth forecasts. Once these data streams were validated, they were used as input data to the HAIMOS forecast. The model was developed using data from several climatologically distinct locations with potential for high solar penetration. In the last year of the project, we conducted a validation campaign according to the guidelines stipulated by the Topic Area 1 project as described in the FOA. This effort brings, for the first time, proven machine-learning methodologies for generating state-of-the-art solar forecasts interweaved with detailed physics-based models for cloud detection, and cloud optical depth forecasts. HAIMOS will generate accurate irradiance probabilistic forecast to assist in reducing solar generation prediction error. Globally optimized solar forecast models are more likely to impact solar energy stakeholders. The goal of this project was to increase the state-of-the-art forecast skill from their present values of 10 to 35%. At the end of the project, we achieved between 30% and 50% forecast skill across a wide range of horizons for both GHI and DNI.

14 SOLAR ENERGY↗

Basaltic Fissure Types on Earth: Suitable Analogs to Evaluate the Origins of Volcanic Terrains on the Moon and Mars?

Basaltic eruptive fissures of the Great Rift and surroundings on the eastern Snake River Plain of Idaho, USA, and selected volcanic features in Hawai`i, Iceland and northern Africa were surveyed for their relevancy as planetary analogs. Evaluated during field investigations and in satellite imagery for structures, physiography, and geologic setting, fissures were categorized into four broad types: (1) simple, monogenetic fissures with obvious volcanic constructs or deposits, (2) monogenetic fissures now obscured by low shields or relatively large cones, (3) polygenetic volcanic rift zones with multiple vents and deposits, and (4) compound regional fissure systems or dike swarms that comprise major rift zones or large volcanic terrains. Using this classification as an initial base, we surveyed imagery of volcanic features for likely fissure vents in two major geologic settings on the Moon: floor-fractured craters (FFCs) and mare and cryptomare provinces. Two major regions on Mars, the volcanic plains around Alba Mons and the greater Tharsis region, were also surveyed for fissure types and volcanic associations of fissure-like features. The planetary surveys suggest that the proposed classification provides a suitable analog starting point to interpret structures associated with fissure systems on the Moon and Mars. With few exceptions, our survey indicates that each of the studied terrains exhibits a dominant fissure type. Type 1 fissures, most with pyroclastic deposits, prevail in lunar FFCs and mare-like regions; whereas type 2 fissures are ubiquitous in the Tharsis region of Mars and a few exist on the Moon as low shields. Type 3 volcanic rift zones are not common on either the Moon or Mars, although they might become evident in future work on chemically evolved terrains. Type 4 fissures are inferred in mare terrains, often represented as the extensions of major linear rille networks or rimae, with possibly complex dike swarms that were buried beneath voluminous mare basalt lava flows. Likewise, numerous flood lavas on Mars are possibly associated with now-obscured or difficult to define type 4 fissure systems.

Scott S. Hughes↗

An Iris Segmentation Algorithm based on Edge Orientation for Off-angle Iris Recognition

Iris recognition is known as one of the most accurate and reliable biometrics. However, the accuracy of iris recognition systems depends on the quality of data capture and is negatively affected by several factors such as angle, occlusion, and dilation. In this paper, we present a segmentation algorithm for off-angle iris images that uses edge detection, edge elimination, edge classification, and ellipse fitting techniques. In our approach, we first detect all candidate edges in the iris image by using the canny edge detector; this collection contains edges from the iris and pupil boundaries as well as eyelash, eyelids, iris texture etc. Edge orientation is used to eliminate the edges that cannot be part of the iris or pupil. Then, we classify the remaining edge points into two sets as pupil edges and iris edges. Finally, we randomly generate subsets of iris and pupil edge points, fit ellipses for each subset, select ellipses with similar parameters, and average to form the resultant ellipses. Based on the results from real experiments, the proposed method shows effectiveness in segmentation for off-angle iris images.

Karakaya, Mahmut↗

Prediction of bacterial E3 ubiquitin ligase effectors using reduced amino acid peptide fingerprinting

Background Although pathogenic Gram-negative bacteria lack their own ubiquitination machinery, they have evolved or acquired virulence effectors that can manipulate the host ubiquitination process through structural and/or functional mimicry of host machinery. Many such effectors have been identified in a wide variety of bacterial pathogens that share little sequence similarity amongst themselves or with eukaryotic ubiquitin E3 ligases. Methods To allow identification of novel bacterial E3 ubiquitin ligase effectors from protein sequences we have developed a machine learning approach, the SVM-based Identification and Evaluation of Virulence Effector Ubiquitin ligases (SIEVE-Ub). We extend the string kernel approach used previously to sequence classification by introducing reduced amino acid (RED) alphabet encoding for protein sequences. Results We found that 14mer peptides with amino acids represented as simply either hydrophobic or hydrophilic provided the best models for discrimination of E3 ligases from other effector proteins with a receiver-operator characteristic area under the curve (AUC) of 0.90. When considering a subset of E3 ubiquitin ligase effectors that do not fall into known sequence based families we found that the AUC was 0.82, demonstrating the effectiveness of our method at identifying novel functional family members. Feature selection was used to identify a parsimonious set of 10 RED peptides that provided good discrimination, and these peptides were found to be located in functionally important regions of the proteins involved in E2 and host target protein binding. Our general approach enables construction of models based on other effector functions. We used SIEVE-Ub to predict nine potential novel E3 ligases from a large set of bacterial genomes. SIEVE-Ub is available for download at https://doi.org/10.6084/m9.figshare.7766984.v1 or https://github.com/biodataganache/SIEVE-Ub for the most current version.

59 BASIC BIOLOGICAL SCIENCES↗

THE GALACTIC O-STAR SPECTROSCOPIC SURVEY (GOSSS). III. 142 ADDITIONAL O-TYPE SYSTEMS

This is the third installment of the Galactic O-Star Spectroscopic Survey (GOSSS), a massive spectroscopic survey of Galactic O stars, based on new homogeneous, high signal-to-noise ratio, R ∼ 2500 digital observations selected from the Galactic O-Star Catalog. In this paper, we present 142 additional stellar systems with O stars from both hemispheres, bringing the total of O-type systems published within the project to 590. Among the new objects, there are 20 new O stars. We also identify 11 new double-lined spectroscopic binaries, 6 of which are of O+O type and 5 of O+B type, and an additional new tripled-lined spectroscopic binary of O+O+B type. We also revise some of the previous GOSSS classifications, present some egregious examples of stars erroneously classified as O-type in the past, introduce the use of luminosity class IV at spectral types O4-O5.5, and adapt the classification scheme to the work of Arias et al.

79 ASTRONOMY AND ASTROPHYSICS↗

Elemental Analyses of Hypervelocity Microparticle Impact Sites on Interplanetary Dust Experiment Sensor Surfaces

The Interplanetary Dust Experiment (IDE) had over 450 electrically active ultra-high purity metal-oxide-silicon impact detectors located on the six primary sides of the Long Duration Exposure Facility (LDEF). Hypervelocity microparticles (approximately 0.2 to approximately 100 micron diameter) that struck the active sensors with enough energy to breakdown the 0.4 or 1.0 micron thick SiO2 insulator layer separating the silicon base (the negative electrode), and the 1000 A thick surface layer of aluminum (the positive electrode) caused electrical discharges that were recorded for the first year of orbit. The high purity Al-SiO2-Si substrates allowed detection of trace (ppm) amounts of hypervelocity impactor residues. After sputtering through a layer of surface contamination, secondary ion mass spectrometry (SIMS) was used to create two-dimensional elemental ion intensity maps of microparticle sites on the IDE sensors. The element intensities in the central craters of the impacts were corrected for relative ion yields and instrumental conditions and then normalized to silicon. The results classification resulted from the particles' origins as 'manmade', 'natural', or 'indeterminate'. The last classification resulted from the presence of too little impactor residue, analytical interference from high background contamination, the lack of information on silicon and aluminum residues, or a combination of these circumstances. Several analytical 'blank' discharges were induced on flight sensors by pressing down on the sensor surface with a pure silicon shard. Analyses of these blank discharges showed that the discharge energy blasts away the layer of surface contamination. Only Si and Al were detected inside the discharge zones, including the central craters, of these features. Thus far, a total of 79 randomly selected microparticle impact sites from the six primary sides of the LDEF were analyzed: 36 from tray C-9 (Leading (ram), or east, side), 18 from tray C-3 (Trailing (wake), or west, side), 12 from tray B-12 (north side), 4 from tray D-6 (south side), 3 from tray H-11 (space end), and 6 from tray G-10 (earth end). Residue from manmade debris was identified in craters on all trays (aluminum oxide particle residues were not detectable on the Al/Si substrates). These results were consistent with the IDE impact record which showed highly variable long term microparticle impact flux rates on the west, space, and Earth sides of the LDEF which could not be ascribed to astronomical variability of micrometeorite density. The IDE record also showed episodic bursts of microparticle impacts on the east, north, and south sides of the satellite, denoting passage through orbital debris clouds or rings.

Simon, C. G.↗

Elemental analyses of hypervelocity microparticle impact sites on Interplanetary Dust Experiment sensor surfaces

The Interplanetary Dust Experiment (IDE) had over 450 electrically active ultra-high purity metal-oxide-silicon impact detectors located on the six primary sides of the Long Duration Exposure Facility (LDEF). Hypervelocity microparticles (approximately 0.2 to approximately 100 micron diameter) that struck the active sensors with enough energy to break down the 0.4 or 1.0 micron thick SIO2 insulator layer separating the silicon base (the negative electrode), and the 1000 A thick surface layer of aluminum (the positive electrode) caused electrical discharges that were recorded for the first year of orbit. The high purity Al-SiO2-Si substrates allowed detection of trace (ppm) amounts of hypervelocity impactor residues. After sputtering through a layer of surface contamination, secondary ion mass spectrometry (SIMS) was used to create two-dimensional elemental ion intensity maps of microparticle impact sites on the IDE sensors. The element intensities in the central craters of the impacts were corrected for relative ion yields and instrumental conditions and then normalized to silicon. The results were used to classify the particles' origins as 'manmade,' 'natural,' or 'indeterminate.' The last classification resulted from the presence of too little impactor residue, analytical interference from high background contamination, the lack of information on silicon and aluminum residues, or a combination of these circumstances. Several analytical 'blank' discharges were induced on flight sensors by pressing down on the sensor surface with a pure silicon shard. Analyses of these blank discharges showed that the discharge energy blasts away the layer of surface contamination. Only Si and Al were detected inside the discharge zones, including the central craters of these features. Thus far a total of 79 randomly selected microparticle impact sites from the six primary sides of the LDEF have been analyzed: 36 from tray C-9 (Leading (ram), or East, side), 18 from tray C-3 (Trailing (wake), or West, side), 12 from tray B-12 (North side), 4 from tray D-6 (South side), 3 from tray H-11 (Space end), and 6 from tray G-10 (Earth end). Residue from manmade debris was identified in craters on all trays. (Aluminum oxide particle residues were not detectable on the Al/Si substrates.) These results were consistent with the IDE impact record which showed highly variable long term microparticle impact flux rates on the West, Space and Earth sides of the LDEF which could not be ascribed to astronomical variability of micrometeorite density. The IDE record also showed episodic bursts of microparticle impacts on the East, North, and South sides of the satellite, denoting passage through orbital debris clouds or rings.

Simon, Charles G.↗

Classification of bacterial plasmid and chromosome derived sequences using machine learning

Plasmids are important genetic elements that facilitate horizonal gene transfer between bacteria and contribute to the spread of virulence and antimicrobial resistance. Most bacterial genome sequences in the public archives exist in draft form with many contigs, making it difficult to determine if a contig is of chromosomal or plasmid origin. Using a training set of contigs comprising 10,584 chromosomes and 10,654 plasmids from the PATRIC database, we evaluated several machine learning models including random forest, logistic regression, XGBoost, and a neural network for their ability to classify chromosomal and plasmid sequences using nucleotide k-mers as features. Based on the methods tested, a neural network model that used nucleotide 6-mers as features that was trained on randomly selected chromosomal and plasmid subsequences 5kb in length achieved the best performance, outperforming existing out-of-the-box methods, with an average accuracy of 89.38% ± 2.16% over a 10-fold cross validation. The model accuracy can be improved to 92.08% by using a voting strategy when classifying holdout sequences. In both plasmids and chromosomes, subsequences encoding functions involved in horizontal gene transfer—including hypothetical proteins, transporters, phage, mobile elements, and CRISPR elements—were most likely to be misclassified by the model. This study provides a straightforward approach for identifying plasmid-encoding sequences in short read assemblies without the need for sequence alignment-based tools.

59 BASIC BIOLOGICAL SCIENCES↗

Radiation Therapy for Primary Cutaneous Anaplastic Large Cell Lymphoma: An International Lymphoma Radiation Oncology Group Multi-institutional Experience

Purpose: To collect response rates of primary cutaneous anaplastic large cell lymphoma, a rare cutaneous T-cell lymphoma, to radiation therapy (RT), and to determine potential prognostic factors predictive of outcome. Methods and Materials: The study was a retrospective analysis of patients with primary cutaneous anaplastic large cell lymphoma who received RT as primary therapy or after surgical excision. Data collected include initial stage of disease, RT modality (electron/photon), total dose, fractionation, response to treatment, and local recurrence. Radiation therapy was delivered at 8 participating International Lymphoma Radiation Oncology Group institutions worldwide. Results: Fifty-six patients met the eligibility criteria, and 63 tumors were treated: head and neck (27%), trunk (14%), upper extremities (27%), and lower extremities (32%). Median tumor size was 2.25 cm (range, 0.6-12 cm). T classification included T1, 40 patients (71%); T2, 12 patients (21%); and T3, 4 patients (7%). The median radiation dose was 35 Gy (range, 6-45 Gy). Complete clinical response (CCR) was achieved in 60 of 63 tumors (95%) and partial response in 3 tumors (5%). After CCR, 1 tumor recurred locally (1.7%) after 36 Gy and 7 months after RT. This was the only patient to die of disease. Conclusions: Primary cutaneous anaplastic large cell lymphoma is a rare, indolent cutaneous lymphoma with a low death rate. This analysis, which was restricted to patients selected for treatment with radiation, indicates that achieving CCR was independent of radiation dose. Because there were too few failures (<2%) for statistical analysis on dose response, 30 Gy seems to be adequate for local control, and even lower doses may suffice.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

TU-F-18C-09: Mammogram Surveillance Using Texture Analysis for Breast Cancer Patients After Radiation Therapy

Purpose: To study the feasibility of applying cancer risk model established from treated patients to predict the risk of recurrence on follow-up mammography after radiation therapy for both ipsilateral and contralateral breast. Methods: An extensive set of textural feature functions was applied to a set of 196 Mammograms from 50 patients. 56 Mammograms from 28 patients were used as training set, 44 mammograms from 22 patients were used as test set and the rest were used for prediction. Feature functions include Histogram, Gradient, Co-Occurrence Matrix, Run-Length Matrix and Wavelet Energy. An optimum subset of the feature functions was selected by Fisher Coefficient (FO) or Mutual Information (MI) (up to top 10 features) or a method combined FO, MI and Principal Component (FMP) (up to top 30 features). One-Nearest Neighbor (1-NN), Linear Discriminant Analysis (LDA) and Nonlinear Discriminant Analysis (NDA) were utilized to build a risk model of breast cancer from the training set of mammograms at the time of diagnosis. The risk model was then used to predict the risk of recurrence from mammogram taken one year and three years after RT. Results: FPM with NDA has the best classification power in classifying the training set of the mammogram with lesions versus those without lesions. The model of FPM with NDA achieved a true positive (TP) rate of 82% compared to 45.5% of using FO with 1-NN. The best false positive (FP) rates were 0% and 3.6% in contra-lateral breast of 1-year and 3-years after RT, and 10.9% in ipsi-lateral breast of 3-years after RT. Conclusion: Texture analysis offers high dimension to differentiate breast tissue in mammogram. Using NDA to classify mammogram with lesion from mammogram without lesion, it can achieve rather high TP and low FP in the surveillance of mammogram for patient with conservative surgery combined RT.

60 APPLIED LIFE SCIENCES↗

Lung Cancer Survival Prediction using Ensemble Data Mining on Seer Data

We analyze the lung cancer data available from the SEER program with the aim of developing accurate survival prediction models for lung cancer. Carefully designed preprocessing steps resulted in removal/modification/splitting of several attributes, and 2 of the 11 derived attributes were found to have significant predictive power. Several supervised classification methods were used on the preprocessed data along with various data mining optimizations and validations. In our experiments, ensemble voting of five decision tree based classifiers and meta-classifiers was found to result in the best prediction performance in terms of accuracy and area under the ROC curve. We have developed an on-line lung cancer outcome calculator for estimating the risk of mortality after 6 months, 9 months, 1 year, 2 year and 5 years of diagnosis, for which a smaller non-redundant subset of 13 attributes was carefully selected using attribute selection techniques, while trying to retain the predictive power of the original set of attributes. Further, ensemble voting models were also created for predicting conditional survival outcome for lung cancer (estimating risk of mortality after 5 years of diagnosis, given that the patient has already survived for a period of time), and included in the calculator. The on-line lung cancer outcome calculator developed as a result of this study is available at http://info.eecs.northwestern.edu:8080/LungCancerOutcomeCalculator/.

Agrawal, Ankit↗

ICESat-2 Photon Classification: Finding Signal Photons in the ATL03 Geolocated Photon Data Product

ICESat-2 carries NASA's next-generation laser altimeter, ATLAS, (Advanced Topographic Laser Altimeter System), designed to measure changes in ice sheet height, sea ice freeboard, and vegetation canopy height. ATLAS contains a photon-counting lidar which transmits green (532-nm) pulses at 10kHz. Each pulse is split into 3 pairs of beams (one strong and one weak). Approximately 1014 photons per pulse travel from ATLAS through the atmosphere to reflect off the Earth's surface. Some return back into the ATLAS telescope where they are recorded. Photons from sunlight and instrument noise at the same wavelength are also recorded. The flight software time tags all photons within a 500m to 6 km range window and generates histograms. Using the histograms, it selects a telemetry window which varies from 20m over flat surfaces to hundreds of meters over rougher terrain. ATL03 contains the time, height (relative to the WGS-84 ellipsoid), latitude and longitude of every photon within the telemetry window. The basic challenge is to determine which of these photons were reflected off the surface. We have developed an algorithm that identifies these signal photons and assigns a confidence level (low, medium, or high) to each signal photon based on the signal to noise ratio. We present an overview of the signal identification algorithm and show the results on actual ICESat-2 data over ice sheet, sea ice, vegetated, and water surfaces. Higher level ATLAS products work with aggregations of the photons in order to determine the ellipsoidal height of the Earth, canopy height and structure, and other quantities of geophysical interest.

Remote Sensing↗

A Data Mining Approach to Predict In Situ Detoxification Potential of Chlorinated Ethenes

Despite advances in physicochemical remediation technologies, in situ bioremediation treatment based on Dehalococcoides mccartyi (Dhc) reductive dechlorination activity remains a cornerstone approach to remedy sites impacted with chlorinated ethenes. Selecting the best remedial strategy is challenging due to uncertainties and complexity associated with biological and geochemical factors influencing Dhc activity. Guidelines based on measurable biogeochemical parameters have been proposed, but contemporary efforts fall short of meaningfully integrating the available information. Extensive groundwater monitoring data sets have been collected for decades, but have not been systematically analyzed and used for developing tools to guide decision-making. In the present study, geochemical and microbial data sets collected from 35 wells at five contaminated sites were used to demonstrate that a data mining prediction model using the classification and regression tree (CART) algorithm can provide improved predictive understanding of a site’s reductive dechlorination potential. The CART model successfully predicted the 3-month-ahead reductive dechlorination potential with 75.8% and 69.5% true positive rate (i.e., sensitivity) for the training set and the test set, respectively. The machine learning algorithm ranked parameters by relative importance for assessing in situ reductive dechlorination potential. The abundance of Dhc 16S rRNA genes, CH 4 , Fe 2+ , NO 3 – , NO 2 – , and SO 4 2– concentrations, total organic carbon (TOC) amounts, and oxidation–reduction potential (ORP) displayed significant correlations (p < 0.01) with dechlorination potential, with NO 3 – , NO 2 – , and Fe 2+ concentrations exhibiting precedence over other parameters. Contrary to prior efforts, the power of data mining approaches lies in the ability to discern synergetic effects between multiple parameters that affect reductive dechlorination activity. Overall, these findings demonstrate that data mining techniques (e.g., machine learning algorithms) effectively utilize groundwater monitoring data to derive predictive understanding of contaminant degradation, and thus have great potential for improving decision-making tools. Finally, a major need for realizing the predictive capabilities of data mining approaches is a curated, open-access, up-to-date and comprehensive collection of biogeochemical groundwater monitoring data.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Characterizing Quantum Classifier Utility in Natural Language Processing Workflows

Quantum Natural Language Processing (QNLP) develops natural language processing (NLP) models for deployment on quantum computers. We explore feature and data prototype selection techniques to address challenges posed by encoding high dimensional features. Our study builds quantum circuit classifiers that includes classical feature pre-processing, quantum embedding and quantum model training. The quantum models are built on 4 or 6 qubits and the quantum neural network (QNN) uses the established bricklayer design. We compare the dependence of model performance (in terms of accuracy and F1 scores) on feature length, embedding gates and parameterized unitary design. We compare the performance of quantum machine learning models to classical convolution neural network model (CNN) on binary and multi-class classification tasks using two datasets of synthetic features and labels. The first is the ECP-CANDLE P3B3 dataset a corpus of synthetically generated cancer pathology reports. The second dataset is extracted from well-known benchmark dataset (MADELON) - features are generated with a combination of informative, repeated and uninformative features. Both datasets are used for binary classification and multi-class classification with 3 classes. We observe robust, accurate performance from all models on the binary classification tasks, but multiclass classification is a challenge for the quantum models-there is a notable decrease in accuracy when using 3 classes. Overall the performance is comparable in terms of recall and accuracy between QNNs and CNNs, even with large datasets. These results provide a point of comparison between quantum and classical models on real-world datasets.

Hamilton, Kathleen↗

Gradient-Based Novelty Detection Boosted by Self-Supervised Binary Classification

Novelty detection aims to automatically identify out-of-distribution (OOD) data, without any prior knowledge of them. It is a critical step in data monitoring, behavior analysis and other applications, helping enable continual learning in the field. Conventional methods of OOD detection perform multi-variate analysis on an ensemble of data or features, and usually resort to the supervision with OOD data to improve the accuracy. In reality, such supervision is impractical as one cannot anticipate the anomalous data. In this paper, we propose a novel, self-supervised approach that does not rely on any pre-defined OOD data: (1) The new method evaluates the Mahalanobis distance of the gradients between the in-distribution and OOD data. (2) It is assisted by a self-supervised binary classifier to guide the label selection to generate the gradients, and maximize the Mahalanobis distance. In the evaluation with multiple datasets, such as CIFAR-10, CIFAR-100, SVHN and TinyImageNet, the proposed approach consistently outperforms state-of-the-art supervised and unsupervised methods in the area under the receiver operating characteristic (AUROC) and area under the precision-recall curve (AUPR) metrics. We further demonstrate that this detector is able to accurately learn one OOD class in continual learning.

Sun, Jingbo↗

Stereotactic Body Radiation Therapy in Recurrent Hepatocellular Carcinoma

Purpose: To examine the safety and efficacy of Cyberknife stereotactic body radiation therapy (SBRT) and its effect on survival in patients of recurrent hepatocellular carcinoma (HCC). Methods and Materials: This was a matched-pair study. From January 2008 to December 2009, 36 patients with 42 lesions of unresectable recurrent HCC were treated with SBRT. The median prescribed dose was 37 Gy (range, 25 to 48 Gy) in 4-5 fractions over 4-5 consecutive working days. Another 138 patients in the historical control group given other or no treatments were selected for matched analyses. Results: The median follow-up time was 14 months for all patients and 20 months for those alive. The 1- and 2-year in-field failure-free rates were 87.6% and 75.1%, respectively. Out-field intrahepatic recurrence was the main cause of failure. The 2-year overall survival (OS) rate was 64.0%, and median time to progression was 8.0 months. In the multivariable analysis of all 174 patients, SBRT (yes vs. no), tumor size ({<=}4 cm vs. >4 cm), recurrent stage (stage IIIB/IV vs. I) and Child-Pugh classification (A vs. B/C) were independent prognostic factors for OS. Matched-pair analysis revealed that patients undergoing SBRT had better OS (2-year OS of 72.6% vs. 42.1%, respectively, p = 0.013). Acute toxicities were mild and tolerable. Conclusion: SBRT is a safe and efficacious modality and appears to be well-tolerated at the dose fractionation we have used, and its use correlates with improved survival in this cohort of patients with recurrent unresectable HCC. Out-field recurrence is the major cause of failure. Further studies of combinations of SBRT and systemic therapies may be reasonable.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Advances in scientific literature mining for interpreting materials characterization

Abstract Using synchrotron light sources, such as the National Synchrotron Light Source II at Brookhaven National Laboratory, scientists in fields as diverse as physics, biology, and materials science, identify the atomic structure, chemical composition, or other important properties of varied specimens. x-ray spectroscopy from light sources is particularly valuable for materials research with vast information available about reference spectra in the scientific literature. However, as the technique is applicable to many science domains, searching for information about select x-ray spectroscopy spectra is impeded by the sheer number of publications. Moreover, useful information about the context of an experiment or figures presented in papers can be buried among the details, which takes time to assess. This work presents a scientific literature mining system that supports data acquisition, information extraction, and user interaction for referencing x-ray spectra identification and spectral interpretation. The goal is to provide efficient access to useful spectral data to researchers who may spend only a few days at a synchrotron light source. With this system, users browse a classification tree for papers arranged according to x-ray spectroscopic methods, chemical elements, and x-ray absorption spectroscopy edges. Relevant figures are extracted with sentences from the paper that explain them, known as ‘figure explanatory text.’ Notably, this system focuses on semantic aspects (logical analysis) to find figure explanatory text using deep contextualized word embeddings techniques and contains an interface to obtain labeled data from domain experts that is used to evaluate and improve the model.

Park, Gilchan (ORCID:0000000201536646)↗

THE RUNAWAYS AND ISOLATED O-TYPE STAR SPECTROSCOPIC SURVEY OF THE SMC (RIOTS4)

We present the Runaways and Isolated O-Type Star Spectroscopic Survey of the SMC (RIOTS4), a spatially complete survey of uniformly selected field OB stars that covers the entire star-forming body of the Small Magellanic Cloud (SMC). Using the IMACS (Inamori-Magellan Areal Camera and Spectrograph) multislit spectrograph and MIKE (Magellan Inamori Kyocera Echelle) echelle spectrograph on the Magellan telescopes, we obtained spectra of 374 early-type field stars that are at least 28 pc from any other OB candidates. We also obtained spectra of an additional 23 field stars in the SMC bar identified from slightly different photometric criteria. Here, we present the observational catalog of stars in the RIOTS4 survey, including spectral classifications and radial velocities. For three multi-slit fields covering 8% of our sample, we carried out monitoring observations over 9–16 epochs to study binarity, finding a spectroscopic, massive binary frequency of at least ∼60% in this subsample. Classical Oe/Be stars represent a large fraction of RIOTS4 (42%), occurring at much higher frequency than in the Galaxy, consistent with expectation at low metallicity. RIOTS4 confirmed a steep upper initial mass function in the field, apparently caused by the inability of the most massive stars to form in the smallest clusters. Our survey also yields evidence for in situ field OB star formation, and properties of field emission-line star populations, including sgB[e] stars and classical Oe/Be stars. We also discuss the radial velocity distribution and its relation to SMC kinematics and runaway stars. RIOTS4 presents a first quantitative characterization of field OB stars in an external galaxy, including the contributions of sparse, but normal, star formation; runaway stars; and candidate isolated star formation.

79 ASTRONOMY AND ASTROPHYSICS↗