Anomalous Readings Research Articles

Healthcare and anomaly detection: using machine learning to predict anomalies in heart rate data

The application of machine learning algorithms to healthcare data can enhance patient care while also reducing healthcare worker cognitive load. These algorithms can be used to detect anomalous physiological readings, potentially leading to expedited emergency response or new knowledge about the development of a health condition. However, while there has been much research conducted in assessing the performance of anomaly detection algorithms on well-known public datasets, there is less conceptual comparison across unsupervised and supervised performance on physiological data. Moreover, while heart rate data are both ubiquitous and noninvasive, there has been little research specifically for anomaly detection of this type of data. Considering that heart rate data are indicative of both potential health complications and an individual’s physical activity, this is a rich source of largely overlooked data. To this end, we employed and evaluated five machine learning algorithms, two of which are unsupervised and the remaining three supervised, in their ability to detect anomalies in heart rate data. These algorithms were then evaluated on real heart rate data. Findings supported the effectiveness of local outlier factor and random forests algorithms in the task of heart rate anomaly detection, as each model generalized well from their training on simulated heart rate data to real world heart rate data. Furthermore, results support that simulated data can help configure algorithms to a degree of performance when real labeled data are not available and that this type of learning might be especially helpful in initial deployment of a system without prior data.

Identification of the Transient Response of a Capacitive Relative Humidity Sensor

This paper presents the results of identifying the transient response of a relative air humidity sensor. It is shown that the capacitive relative humidity sensor is an integrated microprocessor system that performs joint processing of the readings of the absolute humidity sensor and the temperature sensor. A mathematical model of the relative humidity measurement process is proposed. The transient responses of the absolute humidity sensing element and the temperature sensor are specified under the assumption that they correspond to first-order aperiodic factors. Based on the relationships of thermodynamics, Arden Buck equation, and given transient functions, analytical dependencies are obtained to identify the parameters of the transient response of the capacitive relative humidity sensor. The results of several field experiments are processed and analyzed in accordance with the proposed mathematical model. The effect of anomalous sensor readings is revealed and consists in the fact that with simultaneous stepwise changes in humidity and temperature, the sensor records a change in relative humidity with the opposite sign. It has been established that the cause of the anomalous results of relative humidity measurements is the presence of a large difference between the response time constant of the absolute humidity sensing element and that of the temperature sensor. Measures are proposed to prevent anomalous results of relative humidity measurements.

Real-Time Sensor Anomaly Detection and Identification in Automated Vehicles

Connected and automated vehicles (CAVs) are expected to revolutionize the transportation industry, mainly through allowing for a real-time and seamless exchange of information between vehicles and roadside infrastructure. Although connectivity and automation are projected to bring about a vast number of benefits, they can give rise to new challenges in terms of safety, security, and privacy. To navigate roadways, CAVs need to heavily rely on their sensor readings and the information received from other vehicles and roadside units. Hence, anomalous sensor readings caused by either malicious cyber attacks or faulty vehicle sensors can result in disruptive consequences and possibly lead to fatal crashes. As a result, before the mass implementation of CAVs, it is important to develop methodologies that can detect anomalies and identify their sources seamlessly and in real time. In this paper, we develop an anomaly detection approach through combining a deep learning method, namely convolutional neural network (CNN), with a well-established anomaly detection method, and Kalman filtering with a $\chi ^{2}$ -detector, to detect and identify anomalous behavior in CAVs. Our numerical experiments demonstrate that the developed approach can detect anomalies and identify their sources with high accuracy, sensitivity, and F1 score. In addition, this developed approach outperforms the anomaly detection and identification capabilities of both CNNs and Kalman filtering with a $\chi ^{2}$ -detector method alone. It is envisioned that this research will contribute to the development of safer and more resilient CAV systems that implement a holistic view toward intelligent transportation system (ITS) concepts.

Nominal state in (anomalous readings) book by Ibn Khalawayh(370a.h) semantic study

Research in Quarrian readings is considered an imported scope in studying different patterns of using language in argument period ,that is away from necessity of poetry and anonymous texts and knowing the important period in Arabic language history, in addition to know the effect of context in these readings and changes that, generally period Arabic and holy Quran font and types and linguistic derivation and synonymously in using these patterens and their indications.

Pulverization characteristics of coal affected by magmatic intrusion and analysis of the abnormal gas desorption index on drill cuttings

The gas desorption index of drill cuttings is a basic index that measures the initial desorption capacities of coal seams and predicts coal seam gas outbursts. Following a long period of gas drainage in the No.7 coal seam of the No.86 mining area in the Haizi coal mine, the gas desorption index of drill cuttings was still found to be much higher than the threshold value for outburst risks. This abnormal phenomenon led to the present study of the rational selection of test methods and objects in this context. In this study, particle size distribution, pore structure and gas desorption characteristics of coal samples in anomalous areas were analyzed. We found that desorption characteristics are related to particle size and particle size varies in relation to tectonic stress and magmatic intrusions. It appears that the anomalous readings are related to particle size of the coal, resulting from tectonic/magmatic pulverization. Furthermore, measured particle size of drill cuttings is not actually reflective of coal particle size – larger particles may be comprised of multiple smaller particles. The results show that coal samples with particle size <1 mm accounts for 76.3% of total samples and coal samples with particle size >1 mm only accounts for 23.7% of total samples. The porosity and total pore volume increase as the particle size decreases. The specific surface area increases with decreasing pore diameter. Transitional pores and micropores increase the specific surface area of the coal sample considerably. The desorption capacity increases with decreasing particle size. The additional tectonic stress caused by magmatic intrusion has a crushing effect, and 1–3 mm particles used in test were composed of a large amount of smaller particles, eventually resulting in abnormal gas desorption index phenomena. Therefore, we proposed an improved method for measuring the gas desorption index of pulverized coal.

ClariSense+: An enhanced traffic anomaly explanation service using social network feeds

The explosive growth in social networks that publish real-time content begs the question of whether their feeds can complement traditional sensors to achieve augmented sensing capabilities. One such capability is to explain anomalous sensor readings. In our previous conference paper, we built an automated anomaly clarification service, called ClariSense, with the ability to explain sensor anomalies using social network feeds (from Twitter). In this extended work, we present an enhanced anomaly explanation system that augments our base algorithm by considering both (i) the credibility of social feeds and (ii) the spatial locality of detected anomalies. The work is geared specifically for describing small-footprint anomalies, such as vehicular traffic accidents. The original system used information gain to select more informative microblog items to explain physical sensor anomalies. In this paper, we show that significant improvements are achieved in our ability to explain small-footprint anomalies by accounting for information credibility and further discriminating among high-information-gain items according to the size of their spatial footprint. Hence, items that lack sufficient corroboration and items whose spatial footprint in the blogosphere is not specific to the approximate location of the physical anomaly receive less consideration. We briefly demonstrate the workings of such a system by considering a variety of real-world anomalous events, and comparing their causes, as identified by ClariSense+, to ground truth for validation. A more systematic evaluation of this work is done using vehicular traffic anomalies. Specifically, we consider real-time traffic flow feeds shared by the California traffic system. When flow anomalies are detected, our system automatically diagnoses their root cause by correlating the anomaly with feeds on Twitter. For evaluation purposes, the identified cause is then retroactively compared to official traffic and incident reports that we take as ground truth. Results show a great correspondence between our automatically selected explanations and ground-truth data.

ClariSense+: An enhanced traffic anomaly explanation service using social network feeds

The explosive growth in social networks that publish real-time content begs the question of whether their feeds can complement traditional sensors to achieve augmented sensing capabilities. One such capability is to explain anomalous sensor readings. In our previous conference paper, we built an automated anomaly clarification service, called ClariSense, with the ability to explain sensor anomalies using social network feeds (from Twitter). In this extended work, we present an enhanced anomaly explanation system that augments our base algorithm by considering both (i) the credibility of social feeds and (ii) the spatial locality of detected anomalies. The work is geared specifically for describing small-footprint anomalies, such as vehicular traffic accidents. The original system used information gain to select more informative microblog items to explain physical sensor anomalies. In this paper, we show that significant improvements are achieved in our ability to explain small-footprint anomalies by accounting for information credibility and further discriminating among high-information-gain items according to the size of their spatial footprint. Hence, items that lack sufficient corroboration and items whose spatial footprint in the blogosphere is not specific to the approximate location of the physical anomaly receive less consideration. We briefly demonstrate the workings of such a system by considering a variety of real-world anomalous events, and comparing their causes, as identified by ClariSense+, to ground truth for validation. A more systematic evaluation of this work is done using vehicular traffic anomalies. Specifically, we consider real-time traffic flow feeds shared by the California traffic system. When flow anomalies are detected, our system automatically diagnoses their root cause by correlating the anomaly with feeds on Twitter. For evaluation purposes, the identified cause is then retroactively compared to official traffic and incident reports that we take as ground truth. Results show a great correspondence between our automatically selected explanations and ground-truth data.

Insights Arising When Quantum Mechanics is Reappraised From a Biological Perspective

A recently reported ferroelectric phase transition at 72 Kelvin in ice XIc corroborates an observation made as a physics student in 1967. I attributed anomalous readings from a silica helium thermometer suspended in liquid nitrogen to distortion by ice crystallizing on its surface, suggesting a proton ordered tetragonal variant, ice It of cubic ice had changed shape during an ordering phase transition releasing latent energy as 'ice light', wavelength λ ~ 4μ infrared laser light. My research into its consequences reinterprets the axiomatic basis of science, affording simple accounts of familiar topics. Ice light emitted from pools of nitrogen on Earth's poles during a primordial ice age was multiply reflected in cloud and surface ice, creating DNA, including tRNA analogue 'transport DNAs', tDNAs by selectively polymerizing nucleotides in equatorial waters. They constituted H-bond lined pores through proto-cell membranes, polarization by ice light created an electric field propelling ionic substrate-trace element complexes constituting life’s molecular vocabulary inside. 64 tDNA variants orchestrate biochemistry, supplementing trace elements prevents mental and physical maladies. ‘Minions’ arose from comparing cybernetics, psychology and traditional wisdom. 189 flat anti-parallel β-sheet hairpins with alternate neutral and basic amino acid residues bind 1,701 DNA base pairs, forming a nine-coil abacus. Minions probably evolved to pack DNA in chromosomes for accurate replication. 189*18 arrays of proton ordered Hbonds connecting amino acid ω-amines to DNA phosphates control metabolism, are biological clocks and 'chips in the brain'. When sarcomeres of striated muscle contract, they form ½-wave resonant cavities for λ, exemplifying biological energy coupling. Protons accelerated along minion tunnels drive molecular scale nuclear fusion, if coupled they could resolve global warming. Artificial intelligence modeled on minions would satisfy Turing's vision, allowing for personality and compensating the relativity between perception and conception, reinterpreting quantum mechanics, uncertainty and relativity.

Insights Arising When Quantum Mechanics is Reappraised from a Biological Perspective

A recently reported ferroelectric phase transition at 72 Kelvin in ice XIc corroborates an observation made as a physics student in 1967. I attributed anomalous readings from a silica helium thermometer suspended in liquid nitrogen to distortion by ice crystallizing on its surface, suggesting a proton ordered tetragonal variant, ice It of cubic ice had changed shape during an ordering phase transition releasing latent energy as 'ice light', wavelength λ ~ 4 μ infrared laser light. My research into its consequences reinterprets the axiomatic basis of science, affording simple accounts of familiar topics. Ice light emitted from pools of nitrogen on Earth's poles during a primordial ice age was multiply reflected in cloud and surface ice, creating DNA, including tRNA analogue 'transport DNAs', tDNAs by selectively polymerizing nucleotides in equatorial waters. They constituted H-bond lined pores through proto-cell membranes, polarization by ice light created an electric field propelling ionic substrate-trace element complexes constituting life’s molecular vocabulary inside. 64 tDNA variants orchestrate biochemistry, supplementing trace elements prevents mental and physical maladies. ‘Minions’ arose from comparing cybernetics, psychology and traditional wisdom. 189 flat anti-parallel β-sheet hairpins with alternate neutral and basic amino acid residues bind 1,701 DNA base pairs, forming a nine-coil abacus. Minions probably evolved to pack DNA in chromosomes for accurate replication. 189*18 arrays of proton ordered H-bonds connecting amino acid ω-amines to DNA phosphates control metabolism, are biological clocks and 'chips in the brain. When sarcomeres of striated muscle contract, they form ½-wave resonant cavities for λ, exemplifying biological energy coupling. Protons accelerated along minion tunnels drive molecular scale nuclear fusion, if coupled they could resolve global warming. Artificial intelligence modeled on minions would satisfy Turing's vision, allowing for personality and compensating the relativity between perception and conception, reinterpreting quantum mechanics, uncertainty and relativity.

The utility of ground-penetrating radar and its time-dependence in the discovery of clandestine burials

In the field of forensic investigation burial is a relatively common method of hiding a corpse. The location of clandestine graves is, however, a particularly difficult task in which multiple forensic disciplines such as anthropology, botany or archaeology can provide valuable assistance. The use of GPR (ground-penetrating radar) has recently been introduced as a method in the detection of these graves, but what is the true potential of this tool in an operative search scenario?In this study a total of 11 pig carcasses were buried in two wooded areas, each presenting a similar soil composition. The animals were subsequently exhumed at regular intervals, ranging from 2 to 111 weeks, using systematic GPR analysis of the burial sites and archaeological recovery of the subjects that were then autopsied. GPR proved to be useful in recognizing anomalies at the chosen depths of burial and appeared to be dependent on the state of decay of the samples, producing only slight anomalous readings in the presence of skeletal remains: at 92 weeks from burial the difference in signal was weak and at 111 weeks GPR survey offered no helpful information as to burial location. The experiment, in this particular context, determined the technique as being successful in the presence of recent burials, highlighting the need for a multidisciplinary approach in the operative search for buried human remains.

Very short-term earthquake precursors from GPS signal interference based on the 2013 Nantou and Rueisuei earthquakes, Taiwan

We set up a GPS network with 17 Continuous GPS (CGPS) stations in southwestern Taiwan to monitor real-time crustal deformation. We found that systematic perturbations in GPS signals occurred just a few minutes prior to the 2013 Nantou (ML=6.5) and Rueisuei (ML=6.4) earthquakes in Taiwan. The anomalous pseudorange readings were several millimeters higher or lower than those in the background time period. These systematic anomalies were found as a result of interference of GPS L-band signals by electromagnetic emissions (EMs) prior to the mainshocks. The EMs may occur in the form of harmonic or ultra-wide-band radiation and can be generated during the formation of Mode I cracks at the final stage of earthquake nucleation. We estimated the directivity of the likely EM sources by calculating the inner product of the position vector from a GPS station to a given satellite and the vector of anomalous ground motions recorded by the GPS. The results showed that the predominant inner product generally occurred when the satellite was in the direction either toward or away from the epicenter with respect to the GPS network. Our findings suggest that the GPS network may serve as a powerful tool to detect very short-term earthquake precursors and presumably to locate a large earthquake before it occurs. Nevertheless, a direct measurement of EMs at the site of the GPS array is required in future studies to confirm this hypothesis.

Normal seasonal variations for atmospheric radon concentration: a sinusoidal model

Anomalous radon readings in air have been reported before an earthquake activity. However, careful measurements of atmospheric radon concentrations during a normal period are required to identify anomalous variations in a precursor period. In this study, we obtained radon concentration data for 5 years (2003–2007) that can be considered a normal period and compared it with data from the precursory period of 2008 until March 2011, when the 2011 Tohoku-Oki Earthquake occurred. Then, we established a model for seasonal variation by fitting a sinusoidal model to the radon concentration data during the normal period, considering that the seasonal variation was affected by atmospheric turbulence. By determining the amplitude in the sinusoidal model, the normal variation of the radon concentration can be estimated. Thus, the results of this method can be applied to identify anomalous radon variations before an earthquake.

Impact of plasma noise on a direct thrust measurement system

In order to evaluate the accuracy and sensitivity of a pendulum-type thrust measurement system, a linear variable differential transformer (LVDT) and a laser optical displacement sensor have been used simultaneously to determine the displacement resulting from an applied thrust. The LVDT sensor uses an analog interface, whereas the laser sensor uses a digital interface to communicate the displacement readings to the data acquisition equipment. The data collected by both sensors show good agreement for static mass calibrations and validation with a cold gas thruster. However, the data obtained using the LVDT deviate significantly from that of the laser sensor when operating two varieties of plasma thrusters: a radio frequency (RF) driven plasma thruster, and a DC powered plasma thruster. Results establish that even with appropriate shielding and signal filtering the LVDT sensor is subject to plasma noise and radio frequency interactions which result in anomalous thrust readings. Experimental data show that the thrust determined using the LVDT system in a direct current plasma environment and a RF discharge is approximately a factor of three higher than the thrust values obtained using a laser sensor system for the operating conditions investigated. These findings are of significance to the electric propulsion community as LVDT sensors are often utilized in thrust measurement systems and accurate thrust measurement and the reproducibility of thrust data is key to analyzing thruster performance. Methods are proposed to evaluate system susceptibility to plasma noise and an effective filtering scheme presented for DC discharges.

On contextual domain restriction in categorial grammar

Quantifier domain restriction (QDR) and two versions of nominal restriction (NR) are implemented as restrictions that depend on a previously introduced interpreter and interpretation time in a two-dimensional semantic framework on the basis of simple type theory and categorial grammar. Against Stanley (2002) it is argued that a suitable version of QDR can deal with superlatives like tallest. However, it is shown that NR is needed to account for utterances when the speaker intends to convey different restrictions for multiple uses of the same quantifying determiner. We argue that NR generally fares better with such examples but also observe that examples like Every sailor waves at every sailor might be pragmatically anomalous. An account of contextual domain restriction is proposed that (i) excludes these anomalous readings (but it is shown how they could be included), (ii) makes it possible to express different contextual domain restrictions as long-range dependencies on an interpreter and an interpretation time, and (iii) additionally models restrictions based on locative constructions as general mereological constraints introduced by shifting the index.

Anomalous results with the widely used NRPB/SSI-type passive radon dosemeter

In an industrial hall, with large variations of radon concentration within minutes, simultaneous measurements were done with two types of passive radon detectors and an active radon measuring device. The widely used passive radon detector of the National Radiological Protection Board (NRPB) [Health Protection Agency (former NRPB) (HPA)]/Statens strålskyddsinstitut (Swedish Radiation Protection Institute) (SSI) type produced anomalous results, seemingly uncorrelated to the radon concentration which was in the order of hundreds of becquerels per metre, usually underestimating but occasionally overestimating. We tried to reproduce similar exposure characteristics in our laboratory, but failed to reproduce the anomalous readings. We suspected, but could not prove, that the anomalous results were due to the combination of high radon concentration gradients, with pressure-driven air exchange between the inside of the detector holder and the outside atmosphere. Moreover, this theory was at least partly contradicted when we drilled holes in the detector holder. Although of interest, this effect is not likely to have substantially influenced any radon surveys, given the unusual nature of the exposure that caused the effect.

Sensor faults

Various sensor network measurement studies have reported instances of transient faults in sensor readings. In this work, we seek to answer a simple question: How often are such faults observed in real deployments? We focus on three types of transient faults, caused by faulty sensor readings that appear abnormal. To understand the prevalence of such faults, we first explore and characterize four qualitatively different classes of fault detection methods. Rule-based methods leverage domain knowledge to develop heuristic rules for detecting and identifying faults. Estimation methods predict “normal” sensor behavior by leveraging sensor correlations, flagging anomalous sensor readings as faults. Time-series-analysis-based methods start with an a priori model for sensor readings. A sensor measurement is compared against its predicted value computed using time series forecasting to determine if it is faulty. Learning-based methods infer a model for the “normal” sensor readings using training data, and then statistically detect and identify classes of faults. We find that these four classes of methods sit at different points on the accuracy/robustness spectrum. Rule-based methods can be highly accurate, but their accuracy depends critically on the choice of parameters. Learning methods can be cumbersome to train, but can accurately detect and classify faults. Estimation methods are accurate, but cannot classify faults. Time-series-analysis-based methods are more effective for detecting short duration faults than long duration ones, and incur more false positives than the other methods. We apply these techniques to four real-world sensor datasets and find that the prevalence of faults as well as their type varies with datasets. All four methods are qualitatively consistent in identifying sensor faults, lending credence to our observations. Our work is a first step towards automated online fault detection and classification.

Explanation for Anomalous Readings during Monitoring of a Best Management Practice

The USEPA’s Urban Watershed Management Branch (UWMB) has monitored storm-water drainage and best management practices (BMPs) as part of its overall research program. As part of this effort, continuous monitoring equipment was deployed to measure both storm events and periods between storms in a BMP. Of particular concern were electrodes used to measure ammonium ( NH 4+ ) concentrations. During the third deployment of the electrode there were readings of concern for NH 4+ which exceeded expectations for drift or storm event response. The electrode indicated concentrations of NH 4+ exceeded 2 mg/L, with a sustained maximum concentration of 8 mg/L recorded. These observations occurred between December 5, 2003 and December 10, 2003 after snow events. This period was marked by increased measures of conductivity measured which were contrary to typically low conductivity readings especially after precipitation events. Deicers (sodium chloride) spread on the roadway most likely caused increased NH 4+ readings, as...

IKNOS: inference and knowledge in networks of sensors

This paper presents a framework for managing data from sensor of poor quality, with the objective to reduce at the same time the communication load and hence energy consumption. Each node in a wireless sensor network maintains a simple local model of the data it is collecting and sends its parameters to a central location (sink), where it is executed the global monitoring. Local models are used to simulate sensor's readings, minimising the need of communication with sensors and hence the consumption of their battery; they are updated locally, when sensor readings differ excessively from simulated data. At the sink the global model (a Bayesian Network) is learnt on the simulated data. It is used to identify and replace anomalous readings (outliers) that a sensor should have produced and to detect anomalies missed by any single node (when communication with a sensor is interrupted).

Evaluation of Continuous In Situ Monitoring of Soil Changes with Varying Flooding Regimes

To support investigations of flood tolerance occurring at a field‐based research facility, changes in soil volumetric water content, temperature, redox potential, dissolved oxygen content, and pH over the course of flood events were monitored. Electronic sensors connected to dataloggers for continuous monitoring of these parameters were installed, and soil redox potential and pH were also monitored manually for comparison. Soil volumetric water content showed that soils became saturated quickly following inundation. Soil redox potentials revealed a reduction of the soil with inundation; however, stagnant water treatments did not result in lower redox potentials than flowing water treatments. Similarly, dissolved oxygen content was not lower in the stagnant water treatment. The automated and manual systems detected similar trends in redox potential response to flooding; however, redox potentials measured manually were generally higher and significantly different from those obtained with the automated system. Anomalous readings were detected with the automated measurement of soil pH, indicating further need for improvement of this system.

Statistical Algorithms for Designing Geophysical Surveys to Detect UXO Target Areas

The U.S. Department of Defense is in the process of assessing and remediating closed, transferred, and transferring military training ranges across the United States. Many of these sites have areas that are known to contain unexploded ordnance (UXO). Other sites or portions of sites are not expected to contain UXO, but some verification of this expectation using geophysical surveys is needed. Many sites are so large that it is often impractical and/or cost prohibitive to perform surveys over 100% of the site. In such cases, it is particularly important to be explicit about the performance required of the surveys. This article presents the statistical algorithms developed to support the design of geophysical surveys along transects (swaths) to find target areas (TAs) of anomalous geophysical readings that may indicate the presence of UXO. The algorithms described here determine (1) the spacing between transects that should be used for the surveys to achieve a specified probability of traversing the TA, (2) the probability of both traversing and detecting a TA of anomalous geophysical readings when the spatial density of anomalies within the TA is either uniform (unchanging over space) or has a bivariate normal distribution, and (3) the probability that a TA exists when it was not found by surveying along transects. These algorithms have been implemented in the Visual Sample Plan (VSP) software to develop cost-effective transect survey designs that meet performance objectives.