Statistical Outlier Detection (SOD): A computer program for detecting outliers in data
There are no author-identified significant results in this report.
Chhikara, R. S.↗
SEARCH · Engineering Papers
Results for “Detection”
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.
NASA Failure Detection and Prevention Program: Risk Identification and Management Tools - Risk Balancing Profiles (RBP) Defect Detection and Prevention (DDP)
Explore the source record for details and available documents.
risk management failure prevention↗
TPF Planet Detection Testbed: demonstrating deep, stable nulling and planet detection
The design of a testbed being built at the Jet Propulsion Laboratory is described. Simulatiung a dual chopped Bracewell interferometer, the testbed comprises a four beam star and planet source and nulling beam combiner. Since achieving a stable null is of great concern the testbed has many control systems designed to achieve stability of alignment and optical path difference over long periods of time. Comparisons between the testbed and the flight system are drawn and key performance parameters are discussed. The interaction between designs for phaseplate systems that achromatically invert the electric field of one of each pair of the incoming beams to achieve the null and the choice of fringe tracking schemes is also discussed.
interferometry↗
Extraction Instruments to Enable Detection of Origin-Diagnostic Lipids for Life Detection
Explore the source record for details and available documents.
M B Wilhelm↗
A Mass Spectrometry-Machine Learning Approach for Detecting Volatile Organic Compound Emissions for Early Fire Detection
Explore the source record for details and available documents.
Sarah Kingsley↗
Hydrogels for isotopic neutron emitters and antineutrino detection and related detection systems and methods
Scintillation compositions comprising a gadolinium compound and a scintillation compound in a polymer matrix precursor. The scintillation compound comprises strontium diiodide, fac-tris(2-phenylpyridine)iridium (Ir(ppy)3), a quinine compound, or combinations thereof, or 2-(4-biphenylyl)-5 phenyl-1,3,4-oxadiazole (PBD), 2-(4-tert-butylphenyl)-5-(4-phenylphenyl)-1,3,4-oxadiazole (b-PBD), 2,5-diphenyl oxazole (PPO), 1,4-bis(5-phenyloxazol-2-yl) benzene (POPOP), or combinations thereof. Hydrogels comprising the gadolinium compound and scintillation compound in a polymer matrix are also disclosed, as are related systems and methods.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Multi-Dimensional Damage Detection for Surfaces and Structures
Current designs for inflatable or semi-rigidized structures for habitats and space applications use a multiple-layer construction, alternating thin layers with thicker, stronger layers, which produces a layered composite structure that is much better at resisting damage. Even though such composite structures or layered systems are robust, they can still be susceptible to penetration damage. The ability to detect damage to surfaces of inflatable or semi-rigid habitat structures is of great interest to NASA. Damage caused by impacts of foreign objects such as micrometeorites can rupture the shell of these structures, causing loss of critical hardware and/or the life of the crew. While not all impacts will have a catastrophic result, it will be very important to identify and locate areas of the exterior shell that have been damaged by impacts so that repairs (or other provisions) can be made to reduce the probability of shell wall rupture. This disclosure describes a system that will provide real-time data regarding the health of the inflatable shell or rigidized structures, and information related to the location and depth of impact damage. The innovation described here is a method of determining the size, location, and direction of damage in a multilayered structure. In the multi-dimensional damage detection system, layers of two-dimensional thin film detection layers are used to form a layered composite, with non-detection layers separating the detection layers. The non-detection layers may be either thicker or thinner than the detection layers. The thin-film damage detection layers are thin films of materials with a conductive grid or striped pattern. The conductive pattern may be applied by several methods, including printing, plating, sputtering, photolithography, and etching, and can include as many detection layers that are necessary for the structure construction or to afford the detection detail level required. The damage is detected using a detector or sensory system, which may include a time domain reflectometer, resistivity monitoring hardware, or other resistance-based systems. To begin, a layered composite consisting of thin-film damage detection layers separated by non-damage detection layers is fabricated. The damage detection layers are attached to a detector that provides details regarding the physical health of each detection layer individually. If damage occurs to any of the detection layers, a change in the electrical properties of the detection layers damaged occurs, and a response is generated. Real-time analysis of these responses will provide details regarding the depth, location, and size estimation of the damage. Multiple damages can be detected, and the extent (depth) of the damage can be used to generate prognostic information related to the expected lifetime of the layered composite system. The detection system can be fabricated very easily using off-the-shelf equipment, and the detection algorithms can be written and updated (as needed) to provide the level of detail needed based on the system being monitored. Connecting to the thin film detection layers is very easy as well. The truly unique feature of the system is its flexibility; the system can be designed to gather as much (or as little) information as the end user feels necessary. Individual detection layers can be turned on or off as necessary, and algorithms can be used to optimize performance. The system can be used to generate both diagnostic and prognostic information related to the health of layer composite structures, which will be essential if such systems are utilized for space exploration. The technology is also applicable to other in-situ health monitoring systems for structure integrity.
Williams, Martha↗
Final Report on the Detection of Green Monopropellants
In 2014, National Aeronautics and Space Administration (NASA) Kennedy Space Center (KSC) funded a project titled "Familiarization and Detection of Green Monopropellants" utilizing Independent Research and Technology Development (IR&TD) and Center Innovation Fund (CIF) funding. The purpose of the project was to evaluate methods of detection for ammonium dinitramide (ADN) and hydroxylammonium nitrate (HAN). An Engineering Services Contract (ESC) task order was created with the scope of evaluation of several methods of detecting ADN- and HAN-based propellants, as well as development of methods for detection. Detection methods include developed methods such as colorimetric indicating absorbent socks, and commercial-off-the- shelf (COTS) units for ammonia detection. An additional goal of the task order was for ESC to become familiar with ADN's and HAN's material properties and material compatibility. Two approaches were initially investigated as possible methods for the detection of HAN (or AFM315E) and ADN (or LMP-103S). These approaches were colorimetric analysis and instrumentation-based COTS vapor sensors utilization. Initial testing showed that the relatively non-existent vapor pressure of the AF-M315E (of which HAN is a major component) propellant would make the use of COTS sensors difficult for real-time area monitoring of HAN; a small response was detected through the use of active COTS sensors, including the RAE Systems MultiRAE Lite and Drager X-act (registered) 5000, but the levels detected were below the threshold limit value for the toxic gas ammonia. Therefore, a detection system ased upon a colorimetric indicator impregnated into an absorbent material was developed. Preliminary analysis (ESC-245-FDG-001) identified a particularly outstanding candidate as a colorimetric indicator for the detection of the presence of AF-M315E in the form of a Methyl Red (Basic) indicator. Materials impregnated with this indicator exhibit significant color change and the materials are not susceptible to interference from exposure to water or carbon dioxide. The completed detection system for HAN/AF-M315E consists of absorbent socks packed with Fisher Universal Spill Absorbent capable of absorbing and containing any propellant spills that they come into contact with along with indicating wipes. The absorbent socks are also chemically treated with a Methyl Red (Basic) indicator solution to provide the end user with a visual indication that a leak has occurred and proper protective precautions must be undertaken. An added benefit of this detection system is that the absorbent socks should neutralize/absorb any commodity that it comes into contact with (until saturation is reached). Additional adsorbent socks can be deployed until a color change is not seen, indicating that the HAN/AF-M315E contamination has been contained. The indicating wipes provide the user the opportunity to wipe surfaces to determine if there is any HAN/AF-M315E or HAN/AFM315E residue present. The wipes should allow the detection of fuel levels that may be too small to detect with the absorbent socks. The development of a detection system for the ADN/LMP-103S focused on the use of various COTS sensors used as real-time area monitoring devices and personal dosimeters. These COTS based sensor systems were of several different types, including both actively pumped and diffusion-based passive systems, as well as a "rope"-type chemical sensing cable. The results highlighted some of the major differences between the two monopropellants undergoing evaluation. Unlike HAN, ADN (which is the major constituent of LMP-103S) exhibits a much more volatile nature in comparison to AF-M315E. In fact, testing showed that a large percentage of the fuel was lost during the sampling measurement (greater than 10 percent by mass); although this testing cannot tell if the volatile component is the ADN itself or another component of the monopropellant solution. Not surprisingly, all four of the procured vapor-based COTS sensors showed positive results when exposed to solutions of the LMP-103S (ESC-245-FDG-002). The completed detection system for ADN/LMP-103S consists of a combination of two of the tested COTS sensor systems, the RAE Systems MultiRAE Lite and the BW Technologies GasAlert Extreme. These systems are meant to be used in conjunction with one another, which allows for the end-user to have both real-time area monitoring (MultiRAE Lite) as well as a personal dosimeter device (GasAlert Extreme) which can be worn as additional personal protective equipment. An stainless steel extension wand was fabricated and included in the detection system for the MultiRAE Lite to allow for more remote sensing, and connects via the active pumping inlet of the sensor. As stated, the final results of this testing resulted in the production of two "kits" which can be used for the detection of HAN/AF-M315E and ADN/LMP-103s (ESC-245-FDG-003).
Detection↗
Improved Detection System Description and New Method for Accurate Calibration of Micro-Channel Plate Based Instruments and Its Use in the Fast Plasma Investigation on NASA's Magnetospheric MultiScale Mission
The Fast Plasma Investigation (FPI) on NASAs Magnetospheric MultiScale (MMS) mission employs 16 Dual Electron Spectrometers (DESs) and 16 Dual Ion Spectrometers (DISs) with 4 of each type on each of 4 spacecraft to enable fast (30 ms for electrons; 150 ms for ions) and spatially differentiated measurements of the full 3D particle velocity distributions. This approach presents a new and challenging aspect to the calibration and operation of these instruments on ground and in flight. The response uniformity, the reliability of their calibration and the approach to handling any temporal evolution of these calibrated characteristics all assume enhanced importance in this application, where we attempt to understand the meaning of particle distributions within the ion and electron diffusion regions of magnetically reconnecting plasmas. Traditionally, the micro-channel plate (MCP) based detection systems for electrostatic particle spectrometers have been calibrated using the plateau curve technique. In this, a fixed detection threshold is set. The detection system count rate is then measured as a function of MCP voltage to determine the MCP voltage that ensures the count rate has reached a constant value independent of further variation in the MCP voltage. This is achieved when most of the MCP pulse height distribution (PHD) is located at higher values (larger pulses) than the detection system discrimination threshold. This method is adequate in single-channel detection systems and in multi-channel detection systems with very low crosstalk between channels. However, in dense multi-channel systems, it can be inadequate. Furthermore, it fails to fully describe the behavior of the detection system and individually characterize each of its fundamental parameters. To improve this situation, we have developed a detailed phenomenological description of the detection system, its behavior and its signal, crosstalk and noise sources. Based on this, we have devised a new detection system calibration method that enables accurate and repeatable measurement and calibration of MCP gain, MCP efficiency, signal loss due to variation in gain and efficiency, crosstalk from effects both above and below the MCP, noise margin, and stability margin in one single measurement. More precise calibration is highly desirable as the instruments will produce higher quality raw data that will require less post-acquisition data correction using results from in-flight pitch angle distribution measurements and ground calibration measurements. The detection system description and the fundamental concepts of this new calibration method, named threshold scan, will be presented. It will be shown how to derive all the individual detection system parameters and how to choose the optimum detection system operating point. This new method has been successfully applied to achieve a highly accurate calibration of the DESs and DISs of the MMS mission. The practical application of the method will be presented together with the achieved calibration results and their significance. Finally, it will be shown that, with further detailed modeling, this method can be extended for use in flight to achieve and maintain a highly accurate detection system calibration across a large number of instruments during the mission.
detection system↗
Large Interferometer for Exoplanets (Life) I. Improved Exoplanet Detection Yield Estimates for A Large Mid-Infrared Space-Interferometer Mission
Context. One of the long-term goals of exoplanet science is the atmospheric characterization of dozens of small exoplanets in order to understand their diversity and search for habitable worlds and potential biosignatures. Achieving this goal requires a space mission of sufficient scale that can spatially separate the signals from exoplanets and their host stars and thus directly scrutinize the exoplanets and their atmospheres. Aims. We seek to quantify the exoplanet detection performance of a space-based mid-infrared (MIR) nulling interferometer that measures the thermal emission of exoplanets. We study the impact of various parameters and compare the performance with that of large single-aperture mission concepts detecting exoplanets in reflected light. Methods. We have developed an instrument simulator that considers all major astrophysical noise sources and coupled it with Monte Carlo of a synthetic exoplanet population around main-sequence stars within 20 pc. This allows us to quantify the number (and types) of exoplanets that our mission concept could detect over a certain time period. Two different scenarios to distribute the observing time among the stellar targets are discussed and different apertures sizes and wavelength ranges are considered. Results. An interferometer consisting of four 2 m apertures working in the 4–18.5 μ.m wavelength range with a total instrument throughput of 5% could detect up to ≈550 exoplanets with radii between 0.5 and 6 R ⊕ with an integrated S/N ≥ 7. At least ≈160 of the detected exoplanets have radii ≤1.5 R ⊕ . Depending on the observing scenario, ≈25–45 rocky exoplanets (objects with radii between 0.5 and 1.5 R ⊕ ) orbiting within the empirical habitable zone (eHZ) of their host stars are among the detections. With four 3.5 m apertures, the total number of detections can increase to up to ≈770, including ≈60–80 rocky eHZ planets. With four times 1 m apertures, the maximum detection yield is ≈315 exoplanets, including ≤20 rocky eHZ planets. The vast majority of small, temperate exoplanets are detected around M dwarfs. The impact of changing the wavelength range to 3–20 μm or 6–17 μm on the detection yield is negligible. Conclusions. A large space-based MIR nulling interferometer will be able to directly detect hundreds of small, nearby exoplanets, tens of which would be habitable world candidates. In terms of predicted detection yield, such a mission can compete with large single-aperture reflected light missions. Given that we assumed only 2.5 years for the search phase, a significant sub-set of the detected exoplanets can be followed-up in a second mission phase to obtain high SNR thermal emission spectra, leveraging the superior diagnostic power of the MIR wavelength regime compared to shorter wavelengths.
planets and satellites: terrestrial planets↗
Process Anomaly Detection for Sparsely Labeled Events in Nuclear Power Plants
An essential aspect of online monitoring, subtle anomaly detection increases the detection lead time for equipment failure and enables a nuclear power plant (NPP) to mitigate unexpected partial or full outages, resulting in significant cost saving to the plant. Once an anomaly is detected by plant staff, its cause and severity are investigated. Because the vast majority of anomalies require some level of investigation, including some that require time-consuming examination, before they are passed over to the engineering organization for further analysis, plants are often equipped with tools to assist the staff in performing anomaly detection. Those tools operate as a black box and are often based on statistical methods that establish sensor correlations using preconfigured mathematical models and flag correlation deviations as anomalies. Due to the number of anomalies detected at a given NPP on a daily basis, a significant number of flagged anomalies usually await examination for days or weeks. A primary cause of this backlog is that the methods used by the tools generate many false positives. Though this is usually attributed to oversensitive model settings due to very narrow normal operation bands, it can also be associated with the model development being inadequate for the process being monitored, or with missing model inputs that could have explained misclassified positives. The performance of anomaly detection tools impacts their plant acceptance and utilization, especially when the effort to address false positives generated by the tool depletes the value or cost saved by using that tool. Thus, means to advance anomaly detection performance have been investigated by the Department of Energy’s Light Water Reactor Sustainability program. Previous and ongoing efforts have targeted unsupervised machine-learning (ML) methods, which do not require the labeling of any data fed into the ML model. By contrast, in supervised anomaly detection methods, every data point is labeled as either a normal or abnormal process condition, and the model is trained to replicate the classification process. Supervised methods usually outperform unsupervised methods, due to the added value in differentiating normal from anomalous states of the monitored process. An NPP’s corrective action program requires it to track and document, via a dedicated report, the resolution of any issues that occur within the plant. Once created, each report is reviewed by a plant screening committee, and several classifications and decisions are made. Recently, a collaborating NPP developed an artificial intelligence and ML-based classifier to categorize a condition report (CR) into classes that can serve to label the data as normal or anomalous. Applying CRs as labels represents a semi-supervised use case. Semi-supervised ML assumes that labels exist for some data points (i.e., labeled anomalies, in this case) but not for the rest. In this effort, semi-supervised ML methods were used to fuse data from CRs with anomaly detection methods in order to test the hypothesis that partially labeled anomalies would improve the accuracy of the anomaly detection methods. Specifically, two methods were used. The first is the deep Semi-supervised Anomaly Detection (deep SAD) method, which can handle labels ranging from fully unsupervised to fully supervised cases. The second is a newly designed ML method developed specifically for this effort and referred to as the high-order feature (HOF)-based method. To evaluate these two methods in controlled environments, synthetic data generators were developed and used. The first datasets used a spring-mass-damper (SMD) system simulator commonly found in mechanical engineering references. This was used to create two use cases: a one- and a three-mass system. Anomalies were introduced by changing the spring and damper coefficients while the system was actuated by random forces. The second datasets used the commercial Dymola-Modelica software to build a simplified nuclear reactor model. Anomalies were added in the form of corrupted sensor readings and/or control commands. The deep SAD method was tested using the SMD system, while the HOF method was tested using both datasets. Application of the deep SAD semi-supervised ML method demonstrated that labels can generate increased confidence in detecting true anomalies. This helped increase the number of true positives and decrease the number of false negatives—something that would aid in addressing the backlog of possible anomalies. Application of the HOF method demonstrated that labels can aid in down selecting from a candidate set of features to a more optimal subset in order to better differentiate between normal and anomalous conditions.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
Hydrogen Detection Strategies to Support H2@SCALE - The NREL Sensor Laboratory
Hydrogen represents a major pathway to decarbonize and stabilize the national and international energy industry and select manufacturing markets. To facilitate the development of hydrogen markets, the US Department of Energy initiated H2@Scale to bring together stakeholders to advance affordable hydrogen production, transport, storage, and utilization to increase revenue opportunities in multiple energy sectors. One major impediment to hydrogen implementation is cost. To expedite the use of hydrogen in energy and other markets, the United States announced in 2021 the Hydrogen Shot, which seeks to reduce the cost of clean hydrogen by 80% to $1 per 1 kilogram in 1 decade ("1 1 1"). As the cost of hydrogen drops, new applications will emerge that will require unique configurations of existing equipment and infrastructure, and eventually lead to advances in the generation and utilization of hydrogen. As the hydrogen economy expands, sensors and detection methods will need to adapt to changing infrastructure demands to address the primary targets of health & safety, emissions monitoring, and process control. The NREL Sensor Laboratory is playing a pivotal role in advancing the use of hydrogen sensors and detection methodologies in each of these categories to support DOE's mission for safe and efficient utilization in emerging markets. Health & safety monitors are required to ensure that operators and facilities can react to unintended hydrogen releases, either as GH2, LH2, or as a constituent of blends (e.g., natural gas or ammonia). Current detection methodologies focus on safety applications to detect near its lower flammable limit (4 vol %), and typically include point sensors in applications such as fixed or mobile detectors (e.g., personal gas monitors). Methodologies amenable for area detection include acoustic, emerging optical imaging methods, and flame detectors. Comparable detection strategies can be utilized for emissions monitoring and quantization, however few methods can simultaneously cover both low (emissions) and high (health & safety) levels. Deployment of emission level detectors will be required to 1) reduce product loss through small but potentially significant leaks from an environmental or cost perspective, 2) reduce downtime of high demand systems by early identification of eminent system failures (leaks through pump or compressor seals indicative of impending failure), and 3) address potential emission monitoring requirements that may be set by regulating bodies. The first two points should be adopted by industry to reduce the cost-of-goods-sold. The third main category for hydrogen detection relates to process control and may be advantageous for many existing applications. Two main applications are emerging. For example, the purity requirements for hydrogen that is dispensed from refueling systems for hydrogen fuel cell electric vehicles (FCEV) is rigorously regulated by the Standard SAE J2719, which prescribes maximum allowable levels of multiple impurities in the hydrogen fuel and must be verified by a regulatory body. Hydrogen contaminant detectors (HCD) integrated to the fueling station can assure this compliance. HCDs must be able operate in 100% H2 backgrounds and be able to distinguish between multiple contaminants at low ppm to low ppb levels. Secondly, as a strategy to decarbonize the natural gas grid, there are proposals to blend hydrogen with natural gas. This blending will affect transport applications (pipeline infrastructure), stationary combustion systems (turbines), and consumer and commercial appliances. In the short-term, hydrogen levels up to 20% are proposed. Variations in the hydrogen level can have dramatic impact on the combustion process and on the potential response of safety sensors. These mixtures may be regulated so that the concentration at a delivery point must be monitored with high precision. However, routine maintenance may introduce background gases such as ambient air (with water) or maintenance gases (introduced with welding processes or adhesive outgassing.) Therefore, the detection methodology must be robust enough to recover or respond to various contaminants. Several reviews can be found in literature addressing sensing and detection technologies, including their limitations and applications. However, for most applications, limitations can be alleviated by combining various detection techniques either through system integration or implementation of machine learning methods (artificial intelligence). In this presentation, we will discuss several applications, highlight their current approach for hydrogen detection, and suggest detection strategies to supplement their limitations.
ENERGY STORAGE,HYDROGEN↗
Quantifying the Potential of Argon Detection Capabilities for Nuclear Explosion Monitoring
Abstract Current noble gas detection systems for nuclear explosion monitoring are based on the detection of four radioxenon isotopes—Xe-131m, -133, -133m and -135. The data provided by radioxenon detection could be enhanced by other radionuclide signatures such as Ar-37. Activation of Ca-40 in rock by neutrons produces Ar-37, and monitoring for this additional nuclide could help distinguish detections of nuclear explosions from background sources, such as medical isotope production. This work studies the capabilities of a hypothetical argon detection network. A 10 kt explosion was modeled using MCNP and SCALE to determine the inventory of Ar-37 created in a representative granite rock layer, assuming either 0.1, 1 or 10% of the total inventory was released. The Ar-37 inventory was combined with atmospheric transport data from HYSPLIT compiled in a previous study, along with the detection limits of standard Ar-37 detection systems, to determine how many hypothetical monitoring stations would detect Ar-37 from an explosion. This method was repeated for 365 HYSPLIT data sets to create a year’s worth of hypothetical explosions, releases, and detections. The study quantified the average number of detections per release, the number of stations detecting Ar-37, and the possibility of detecting Ar-37 in coincidence with xenon.
37Ar↗
Automatic Detection and Classification of Radio Galaxy Images by Deep Learning
Abstract Surveys conducted by radio astronomy observatories, such as SKA, MeerKAT, Very Large Array, and ASKAP, have generated massive astronomical images containing radio galaxies (RGs). This generation of massive RG images has imposed strict requirements on the detection and classification of RGs and makes manual classification and detection increasingly difficult, even impossible. Rapid classification and detection of images of different types of RGs help astronomers make full use of the observed astronomical image data for further processing and analysis. The classification of FRI and FRII is relatively easy, and there are more studies and literature on them at present, but FR0 and FRI are similar, so it is difficult to distinguish them. It poses a greater challenge to image processing. At present, deep learning has made breakthrough progress in the field of image analysis and processing and has preliminary applications in astronomical data processing. Compared with classification algorithms that can only classify galaxies, object detection algorithms that can locate and classify RGs simultaneously are preferred. In target detection algorithms, YOLOv5 has outstanding advantages in the classification and positioning of small targets. Therefore, we propose a deep-learning method based on an improved YOLOv5 object detection model that makes full use of multisource data, combining FIRST radio with SDSS optical image data, and realizes the automatic detection of FR0, FRI, and FRII RGs. The innovation of our work is that on the basis of the original YOLOv5 object detection model, we introduce the SE Net attention mechanism, increase the number of preset anchors, adjust the network structure of the feature pyramid, and modify the network structure, thereby allowing our model to demonstrate galaxy classification and position detection effects. Our improved model produces satisfactory results, as evidenced by experiments. Overall, the mean average precision (mAP@0.5) of our improved model on the test set reaches 89.4%, which can determine the position (R.A. and decl.) and automatically detect and classify FR0s, FRIs, and FRIIs. Our work contributes to astronomy because it allows astronomers to locate FR0, FRI, and FRII galaxies in a relatively short time and can be further combined with other astronomically generated data to study the properties of these galaxies. The target detection model can also help astronomers find FR0s, FRIs, and FRIIs in future surveys and build a large-scale star RG catalog. Moreover, our work is also useful for the detection of other types of galaxies.
Astronomy & Astrophysics↗
GEDet: Detecting Erroneous Nodes with A Few Examples
Detecting nodes with erroneous values in real-world graphs re- mains challenging due to the lack of examples and various error scenarios. We demonstrate GEDet, an error detection engine that can detect erroneous nodes in graphs with a few examples. The GEDet framework tackles error detection as a few-shot node classification problem. We invite the attendees to experience the following unique features. (1) Few-shot detection. Users only need to provide a few examples of erroneous nodes to perform error detection with GEDet. GEDet achieves desirable accuracy with (a) a graph augmentation module, which automatically generates synthetic examples to learn the classifier, and (b) an adversarial detection module, which improves classifiers to better distinguish erroneous nodes from both cleaned nodes and synthetic examples. We show that GEDet significantly improves the state-of-the-art error detection methods. (2) Diverse error scenarios. GEDet profiles data errors with a built-in library of transformation functions from correct values to errors. Users can also easily “plug in” new error types or examples. (3) User-centric detection. GEDet supports (a) an active learning mode to engage users to verify detected results, and adapts the error detection process accordingly; and (b) visual interfaces to interpret and track detected errors.
Guan, Sheng↗
Coherent Detection of High-Rate Optical PPM Signals
A method of coherent detection of high-rate pulse-position modulation (PPM) on a received laser beam has been conceived as a means of reducing the deleterious effects of noise and atmospheric turbulence in free-space optical communication using focal-plane detector array technologies. In comparison with a receiver based on direct detection of the intensity modulation of a PPM signal, a receiver based on the present method of coherent detection performs well at much higher background levels. In principle, the coherent-detection receiver can exhibit quantum-limited performance despite atmospheric turbulence. The key components of such a receiver include standard receiver optics, a laser that serves as a local oscillator, a focal-plane array of photodetectors, and a signal-processing and data-acquisition assembly needed to sample the focal-plane fields and reconstruct the pulsed signal prior to detection. The received PPM-modulated laser beam and the local-oscillator beam are focused onto the photodetector array, where they are mixed in the detection process. The two lasers are of the same or nearly the same frequency. If the two lasers are of different frequencies, then the coherent detection process is characterized as heterodyne and, using traditional heterodyne-detection terminology, the difference between the two laser frequencies is denoted the intermediate frequency (IF). If the two laser beams are of the same frequency and remain aligned in phase, then the coherent detection process is characterized as homodyne (essentially, heterodyne detection at zero IF). As a result of the inherent squaring operation of each photodetector, the output current includes an IF component that contains the signal modulation. The amplitude of the IF component is proportional to the product of the local-oscillator signal amplitude and the PPM signal amplitude. Hence, by using a sufficiently strong local-oscillator signal, one can make the PPM-modulated IF signal strong enough to overcome thermal noise in the receiver circuits: this is what makes it possible to achieve near-quantum-limited detection in the presence of strong background. Following quantum-limited coherent detection, the outputs of the individual photodetectors are automatically aligned in phase by use of one or more adaptive array compensation algorithms [e.g., the least-mean-square (LMS) algorithm]. Then the outputs are combined and the resulting signal is processed to extract the high-rate information, as though the PPM signal were received by a single photodetector. In a continuing series of experiments to test this method (see Fig. 1), the local oscillator has a wavelength of 1,064 nm, and another laser is used as a signal transmitter at a slightly different wavelength to establish an IF of about 6 MHz. There are 16 photodetectors in a 4 4 focal-plane array; the detector outputs are digitized at a sampling rate of 25 MHz, and the signals in digital form are combined by use of the LMS algorithm. Convergence of the adaptive combining algorithm in the presence of simulated atmospheric turbulence for optical PPM signals has already been demonstrated in the laboratory; the combined output is shown in Fig. 2(a), and Fig. 2(b) shows the behavior of the phase of the combining weights as a function of time (or samples). We observe that the phase of the weights has a sawtooth shape due to the continuously changing phase in the down-converted output, which is not exactly at zero frequency. Detailed performance analysis of this coherent free-space optical communication system in the presence of simulated atmospheric turbulence is currently under way.
Vilnrotter, Victor↗
Assessing the Effectiveness of Generalized Likelihood Ratio Test Detector Schemes in Seismic Event Detection and the Avoidance of Nontarget Signals
Cross-correlation techniques have played a long-standing and pivotal role in seismic event monitoring. However, the performance of correlation-based detectors is challenged by nuisance seismicity, or nontarget signals. Such detections are a problem when the mission is to automatically map events to the correct source region. Using aftershocks of the 2014 $M_w$ 6.0 South Napa, California, earthquake, we demonstrate the effectiveness of utilizing a dynamic correlation processor framework in a generalized likelihood ratio test (GLRT) detector configuration to minimize nontarget detections. A GLRT maximizes a detection statistic with respect to one or more unknown parameters. In this case, the detection statistic is a template signal match against the waveform in a window sliding over a data stream, and the unknown parameter is an index variable indicating group membership of the template event or events. Detected events are assigned to the event group that yields the largest detection statistic. In this work, our results show that a GLRT detector will outperform a suite of independently operating correlation and subspace detectors in terms of having a lower nontarget detection rate at a given missed detection rate. We also show that a GLRT detector composed of a few high-rank subspace detectors has a slightly higher nontarget detection rate, but a significantly lower missed detection rate, than a GLRT detector composed of many low-rank subspace detectors. The high-rank GLRT configuration produced impressive results even with marginal data (single channel, single station, and very low time bandwidth product), which bodes well for the utility of building efficient aftershock classification systems and global monitoring systems at larger scales. However, future work is required to assess performance at the regional scale and to assess the performance of the system at detecting target events not used in the detector template creation.
58 GEOSCIENCES↗
Improving radiograph analysis throughput through transfer learning and object detection
SIGN Fracture Care International partners with surgeons in low-resource hospitals worldwide to provide access to effective orthopedic care by donating educational materials and innovatively designed surgical implants. Over two decades, SIGN’s Online Surgical Database (SOSD) has grown to contain over 500,000 medical images, with radiographs holding the majority share. One challenge in working with hospitals worldwide is that both the radiographs uploaded to the SOSD and the data entry accompanying the uploads vary in quality. To improve the accuracy of data in the SOSD, we trained a model to detect surgical implants in radiographs. We first developed a tool to automatically detect radiographs, then trained an object detection model to determine the number and placement of surgical implants visible in the radiograph. Active learning was used to generate a training set containing 2,510 radiographs with screws, nails, and plates labeled by bounding boxes. Training a model to simultaneously recognize all three classes of implants gave a low average precision (AP) for the plate class, likely due to the low number of plate instances in our training set and the large variety of surgical plates used by SIGN-partnered surgeons. Applying standard image augmentation techniques to increase the plate count in our training set did not appreciably increase the AP of plate detection. To improve plate detection, we redrew the bounding boxes to account for correlations between the screw and plate classes. Training one model to detect nails and screws and a separate model to detect plates increased the AP of plate detection by 78.8 percentage points. The AP of each class was 80.7% for screws, 93.6% for nails, and 92.6% for plates; meanwhile, the sensitivity was 92% for screws, 86% for nails, and 81% for plates. We show that object detection methods can be used to detect surgical implants in radiographs of varying quality; however, the detection ability is dependent on the type of implant, and some implants, in our case plates, must be treated differently than others. Such tools can improve the throughput of radiograph analysis, assisting physicians and surgeons with the treatment of bone fractures.
60 APPLIED LIFE SCIENCES↗