- Hanson Center for Space Sciences
University of Texas at Dallas
MS WT15, 800 W Campbell Rd
Richardson TX 75080
USA
David J Lary
University of Texas at Dallas, Physics, Faculty Member
Abstract : The hypothesis of this study is that a suite of remote sensing data products on atmospheric aerosols used in their meteorological context and processed by machine learning can provide a daily estimate of the global PM 2.5... more
Abstract : The hypothesis of this study is that a suite of remote sensing data products on atmospheric aerosols used in their meteorological context and processed by machine learning can provide a daily estimate of the global PM 2.5 abundance. This information is of considerable value to Global Health Surveillance (GHS), providing a capability to routinely estimate troop deployment exposure to elevated levels of particulate matter (PM) globally, significantly contributing to DoD-wide force health protection initiatives. We have exceeded our promised goal and have provided a daily global estimate of the PM2.5 distribution from February 2000 up through the present (we promised from 2006-present).
Research Interests:
Research Interests:
Research Interests:
Research Interests:
ABSTRACT
Research Interests: Atmospheric Chemistry, Kinetics, Data Assimilation, Numerical Modelling, Acid Rain, and 11 moreControl system, Earth Observation, Automatic Parallelization, Information Content, Dynamic Panel Data, Look Ahead, Automatic code generation, Data Processing, Ozone layer depletion, Metabolic pathway, and Dynamic Adaptation
Research Interests:
Research Interests:
ABSTRACT BigData should be a key component of a holistic approach to public health. There is increasing awareness that health is shaped by more than health care, but the exact causal pathway that links health to health behaviors,... more
ABSTRACT BigData should be a key component of a holistic approach to public health. There is increasing awareness that health is shaped by more than health care, but the exact causal pathway that links health to health behaviors, socioeconomic conditions, and environmental conditions has been inadequately explored by conventional epidemiologic methods. Existing knowledge and conventional research tools are often insufficient to predict a priori how various environmental, social, psychological, behavioral, and biological factors are interrelated and change over time. Human health is an interdependent multifaceted system. The quantity of data that is now available through new technologies requires different analytic methods and approaches. An exciting new era is dawning where we are using these valuable data together (fully multi- variately) with computational techniques such as machine learning to provide insights for integrative health in the areas of methodology for patient care, scientific discovery, decision support, and policy formulation. This session will showcase new advances for those who would like to leverage the computational BigData revolution for integrative health insight generation and describe some areas of exciting future development. Upon completion attendees will have an appreciation of the tremendous value of using the methodology of BigData and Machine Learning for Integrative Health. This is timely as many Public Health professionals may not be familiar with these tools.
Research Interests:
s of Papers Presented at the WMO-IGA C Conference on the Measurement and Assessment of A tmospheric Composition Change, Beij ing, China, 91 3 October, 1 995 , WMO/TD-No. 7 1 0,
Research Interests:
CMOS (Complementary Metal Oxide Semiconductor) transmitter and receiver circuits for rotational spectroscopy are demonstrated. The IC’s implemented in 65-nm CMOS consist of a 208-252 GHz transmitter and a 225-280 GHz receiver. Use of CMOS... more
CMOS (Complementary Metal Oxide Semiconductor) transmitter and receiver circuits for rotational spectroscopy are demonstrated. The IC’s implemented in 65-nm CMOS consist of a 208-252 GHz transmitter and a 225-280 GHz receiver. Use of CMOS electronics can reduce the cost of electronics for rotational spectrometer application from over $50k to less than $1k. The receiver (RX) includes an on-chip antenna for air-to-chip interface, a 2nd order sub-harmonic down-conversion mixer, a low noise IF amplifier and an amplitude detector. The transmitter (TX) includes an on-chip antenna for chip-to-air interface, Fractional-N synthesizer with a frequency step less than 1 kHz with a built-in frequency shift keying circuit as well as a frequency up-converter to generate the signal at the RF. The integrated circuits were assembled into a rotational spectrometer and utilized to detect numerous gases including Ethanol and Acetone in human breath. It is the first ever demonstration of spectroscopy on pure gases as well as breath using CMOS circuits, and this work paves the way toward a more compact, affordable and efficient rotational spectroscopy system.
Research Interests:
Research Interests:
Neural networks are non-linear non-parametric learning algorithms that are universal approximators. They have proved very useful to us in a variety of applications (Lary et al., 2004, 2007a,b), from the acceleration of expensive code... more
Neural networks are non-linear non-parametric learning algorithms that are universal approximators. They have proved very useful to us in a variety of applications (Lary et al., 2004, 2007a,b), from the acceleration of expensive code elements to learning the cross-calibration between large earth observing datasets including atmospheric composition, aerosol optical depth, and vegetation indices. We have been using a variety of networks including feed-forward multi-layer perceptron networks trained with the Levenberg-Marquardt algorithm, and neuro-fuzzy networks. The success of the neural networks largely depends on two factors. First, having a training dataset that adequately spans the parameter space. Second, including the variables that explain the variance in the dataset. If these two criteria are met then the neural networks give excellent results as they are universal approximators. We present several examples of neural network cross-calibration.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
ABSTRACT Assimilation of trace gas observations into atmospheric chemical models is becoming more common as a means to make better use of observations and to improve the models. However, this task is not straightforward. First, there is... more
ABSTRACT Assimilation of trace gas observations into atmospheric chemical models is becoming more common as a means to make better use of observations and to improve the models. However, this task is not straightforward. First, there is the technical challenge of combining an assimilation scheme with a full chemistry 3D model. Second, there is a problem of applying the computationally expensive model to real atmospheric problems in a useful way. In this talk we will describe how we have coupled a sequential assimilation scheme to the SLIMCAT 3D chemical transport model. The sequential assimilation scheme is computationally efficient and can be used for long (multiannual) studies. The scheme implemented can assimilate several species simultaneously and care has been taken to ensure that the assimilation procedure conserves family abundances and the compact correlations which exist among stratospheric long-lived tracers. Results will be shown from the assimilation of HALOE data for 1992. Assimilation of these sparse profiles improves the global distribution of long-lived tracers (e.g. CH4) and, via the conservation of tracer correlations, also improves the distribution of species which are not directly observed (e.g. N2O). The application of the assimilation model to outstanding problems with 3D models will be discussed.
Research Interests:
ABSTRACT
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Black pod rot, caused by Phytophthora palmivora, is a devastating disease of Theobroma cacao L. (cacao) leading to huge losses for farmers and limiting chocolate industry supplies. To understand resistance responses of cacao leaves to P.... more
Black pod rot, caused by Phytophthora palmivora, is a devastating disease of Theobroma cacao L. (cacao) leading to huge losses for farmers and limiting chocolate industry supplies. To understand resistance responses of cacao leaves to P. palmivora, Stage 2 leaves of genotypes Imperial College Selection 1 (ICS1), Colección Castro Naranjal 51 (CCN51), and Pound7 were inoculated with zoospores and monitored for symptoms up to 48 h. Pound7 consistently showed less necrosis than ICS1 and CCN51 48 h after inoculation. RNA-Seq was carried out on samples 24 h post inoculation. A total of 24,672 expressed cacao genes were identified, and 2,521 transcripts showed induction in at least one P. palmivora-treated genotype compared to controls. There were 115 genes induced in the P. palmivora-treated samples in all three genotypes. Many of the differentially expressed genes were components of KEGG pathways important in plant defense signal perception (the plant MAPK signaling pathway, plant hormon...
Research Interests:
Remote sensing imagery, such as that provided by the United States Geological Survey (USGS) Landsat satellites, has been widely used to study environmental protection, hazard analysis, and urban planning for decades. Clouds are a constant... more
Remote sensing imagery, such as that provided by the United States Geological Survey (USGS) Landsat satellites, has been widely used to study environmental protection, hazard analysis, and urban planning for decades. Clouds are a constant challenge for such imagery and, if not handled correctly, can cause a variety of issues for a wide range of remote sensing analyses. Typically, cloud mask algorithms use the entire image; in this study we present an ensemble of different pixel-based approaches to cloud pixel modeling. Based on four training subsets with a selection of different input features, 12 machine learning models were created. We evaluated these models using the cropped LC8-Biome cloud validation dataset. As a comparison, Fmask was also applied to the cropped scene Biome dataset. One goal of this research is to explore a machine learning modeling approach that uses as small a training data sample as possible but still provides an accurate model. Overall, the model trained on...
Research Interests:
The human body is an incredible and complex sensing system. Environmental factors trigger a wide range of automatic neurophysiological responses. Biometric sensors can capture these responses in real time, providing clues about the... more
The human body is an incredible and complex sensing system. Environmental factors trigger a wide range of automatic neurophysiological responses. Biometric sensors can capture these responses in real time, providing clues about the underlying biophysical mechanisms. In this prototype study, we demonstrate an experimental paradigm to holistically capture and evaluate the interactions between an environmental context and physiological markers of an individual operating that environment. A cyclist equipped with a biometric sensing suite is followed by an environmental survey vehicle during outdoor bike rides. The interactions between environment and physiology are then evaluated though the development of empirical machine learning models, which estimate particulate matter concentrations from biometric variables alone. Here, we show biometric variables can be used to accurately estimate particulate matter concentrations at ultra-fine spatial scales with high fidelity (r2 = 0.91) and tha...
Research Interests:
Electroencephalography (EEG) is a brain imaging technique in which electrodes are placed on the scalp. EEG signals are commonly decomposed into frequency bands called delta, theta, alpha, and beta. While these bands have been shown to be... more
Electroencephalography (EEG) is a brain imaging technique in which electrodes are placed on the scalp. EEG signals are commonly decomposed into frequency bands called delta, theta, alpha, and beta. While these bands have been shown to be useful for characterizing various brain states, their utility as a one-size-fits-all analysis tool remains unclear. The goal of this work is to outline an objective strategy for discovering optimal EEG bands based on signal power spectra. A two-step data-driven methodology is presented for objectively determining the best EEG bands for a given dataset. First, a decision tree is used to estimate the optimal frequency band boundaries for reproducing the signal’s power spectrum for a predetermined number of bands. The optimal number of bands is then determined using an Akaike Information Criterion (AIC)-inspired quality score that balances goodness-of-fit with a small band count. This data-driven approach led to better characterization of the underlyin...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Q3 Juan-Pablo Ramirez-Paredes Department of Electrical Engineering, University of Texas at Dallas, 800 West Campbell Road, Richardson, Texas 75080 e-mail: pablo.ramirez@utdallas.edu David Lary Department of Physics, The University of... more
Q3 Juan-Pablo Ramirez-Paredes Department of Electrical Engineering, University of Texas at Dallas, 800 West Campbell Road, Richardson, Texas 75080 e-mail: pablo.ramirez@utdallas.edu David Lary Department of Physics, The University of Texas at Dallas, 800 West Campbell Road, Richardson, Texas 75080 e-mail: david.lary@utdallas.edu Nicholas Gans Department of Electrical Engineering, University of Texas at Dallas, 800 West Campbell Road, Richardson, Texas 75080
Research Interests:
In this research, we present data‐driven forecasting of ionospheric total electron content (TEC) using the Long‐Short Term Memory (LSTM) deep recurrent neural network method. The random forest machine learning method was used to perform a... more
In this research, we present data‐driven forecasting of ionospheric total electron content (TEC) using the Long‐Short Term Memory (LSTM) deep recurrent neural network method. The random forest machine learning method was used to perform a regression analysis and estimate the variable importance of the input parameters. The input data are obtained from satellite and ground based measurements characterizing the solar‐terrestrial environment. We estimate the relative importance of 34 different parameters, including the solar flux, solar wind density, and speed the three components of interplanetary magnetic field, Lyman‐alpha, the Kp, Dst, and Polar Cap (PC) indices. The TEC measurements are taken with 15‐s cadence from an equatorial GPS station located at Bogota, Columbia (4.7110° N, 74.0721° W). The 2008–2017 data set, including the top five parameters estimated using the random forest, is used for training the machine learning models, and the 2018 data set is used for independent te...
Research Interests:
: The hypothesis of this study is that a suite of remote sensing data products on atmospheric aerosols used in their meteorological context and processed by machine learning can provide a daily estimate of the global PM 2.5 abundance.... more
: The hypothesis of this study is that a suite of remote sensing data products on atmospheric aerosols used in their meteorological context and processed by machine learning can provide a daily estimate of the global PM 2.5 abundance. This information is of considerable value to Global Health Surveillance (GHS), providing a capability to routinely estimate troop deployment exposure to elevated levels of particulate matter (PM) globally, significantly contributing to DoD-wide force health protection initiatives. We have exceeded our promised goal and have provided a daily global estimate of the PM2.5 distribution from February 2000 up through the present (we promised from 2006-present).
Research Interests:
Electrochemistry of surface‐bound molecules is of high importance for numerous electronic and sensor applications. Extracting the electron transfer rate is beneficial for understanding surface‐bound processes, but it requires experimental... more
Electrochemistry of surface‐bound molecules is of high importance for numerous electronic and sensor applications. Extracting the electron transfer rate is beneficial for understanding surface‐bound processes, but it requires experimental or computational rigor. We evaluate methods to determine electron transfer rates from large voltammetry sets from experiments via machine learning using decision tree ensembles, neural networks, and gaussian process regression models. We applied these to reproduce computational measures of electron transfer rates modeled by first principles. The best machine learning models were a random forest with 80 decision trees and a neural network with Bayesian regularization, producing root mean squared errors of 0.37 and 0.49 s−1, respectively, corresponding to mean percent errors of 0.38 % and 0.52 %, respectively. This work establishes machine learning methods for rapidly acquiring electron transfer rates across large datasets for widespread applications.
Research Interests:
ABSTRACT There is great utility in having a flexible and automated objective observation direction system for the decadal survey missions and beyond. Such a system allows us to optimize the observations made by suite of sensors to address... more
ABSTRACT There is great utility in having a flexible and automated objective observation direction system for the decadal survey missions and beyond. Such a system allows us to optimize the observations made by suite of sensors to address specific goals from long term monitoring to rapid response. We have developed such a prototype using a network of communicating software elements to control a heterogeneous network of sensor systems, which can have multiple modes and flexible viewing geometries. Our system makes sensor systems intelligent and situationally aware. Together they form a sensor web of multiple sensors working together and capable of automated target selection, i.e. the sensors ``know'' where they are, what they are able to observe, what targets and with what priorities they should observe. This system is implemented in three components. The first component is a Sensor Web simulator. The Sensor Web simulator describes the capabilities and locations of each sensor as a function of time, whether they are orbital, sub-orbital, or ground based. The simulator has been implemented using AGIs Satellite Tool Kit (STK). STK makes it easy to analyze and visualize optimal solutions for complex space scenarios, and perform complex analysis of land, sea, air, space assets, and shares results in one integrated solution. The second component is target scheduler that was implemented with STK Scheduler. STK Scheduler is powered by a scheduling engine that finds better solutions in a shorter amount of time than traditional heuristic algorithms. The global search algorithm within this engine is based on neural network technology that is capable of finding solutions to larger and more complex problems and maximizing the value of limited resources. The third component is a modeling and data assimilation system. It provides situational awareness by supplying the time evolution of uncertainty and information content metrics that are used to tell us what we need to observe and the priority we should give to the observations. A prototype of this component was implemented with AutoChem. AutoChem is NASA release software constituting an automatic code generation, symbolic differentiator, analysis, documentation, and web site creation tool for atmospheric chemical modeling and data assimilation. Its model is explicit and uses an adaptive time-step, error monitoring time integration scheme for stiff systems of equations. AutoChem was the first model to ever have the facility to perform 4D-Var data assimilation and Kalman filter. The project developed a control system with three main accomplishments. First, fully multivariate observational and theoretical information with associated uncertainties was combined using a full Kalman filter data assimilation system. Second, an optimal distribution of the computations and of data queries was achieved by utilizing high performance computers/load balancing and a set of automatically mirrored databases. Third, inter-instrument bias correction was performed using machine learning. The PI for this project was Dr. David Lary of the UMBC Joint Center for Earth Systems Technology at NASA/Goddard Space Flight Center.