The novel coronavirus (COVID-19) epidemic is generating significant social, economic, and health impacts and has highlighted the importance of real-time analysis of the spatio-temporal dynamics of emerging infectious diseases. COVID-19, which emerged out of the city of Wuhan in China in December 2019 is now spreading in multiple countries. It is particularly concerning that the case fatality rate appears to be higher for the novel coronavirus than for seasonal influenza, and especially so for older populations and those with prior health conditions such as cardiovascular disease and diabetes. Any plan for stopping the epidemic must be based on a quantitative understanding of the proportion of the at-risk population that needs to be protected by effective control measures in order for transmission to decline sufficiently and quickly enough for the epidemic to end. Different data collection and testing modalities and strategies available to help calibrate transmission models and predict the spread/severity of a disease, have variable costs, response times, and accuracies. In this Rapid Response Research (RAPID) project, the team will examine the problem of establishing optimal practices for rapid testing for the novel coronavirus. The result will be the Rapid Testing for Epidemic Modeling (RTEM), which will translate into science-based predictions of the COVID-19 epidemic’s characteristics, including the duration and overall size, and help the global efforts to combat the disease. The RTEM will fill an important gap in data-driven decision making during the COVID-19 epidemic and, thus, will enable services with significant national economic and health impact. The educational impact of the project will be on mentoring of post-doctoral and PhD researchers and on curricula by incorporating research challenges and outcomes into existing undergraduate and graduate classes.

Computational models for the spatio-temporal dynamics of emerging infectious diseases and data- and model-driven computer simulations for disease spreading are increasingly critical in predicting geo-temporal evolution of epidemics as well as designing, activating, and adapting practices for controlling epidemics. In this project, the researchers tackle a Rapid Testing for Epidemic Modeling (RTEM) problem: Given a partially known target disease model and a set of testing modalities (from surveys to surveillance testing at known disease hotspots), with varying costs, accuracies, and observational delays, what is the best rapid testing strategy that would help recover the underlying disease model? Several scientific questions arise: What is the value of testing? Should only sick people be tested for virus detection? What level of resources should be devoted to the development of highly accurate tests (low false positives, low false negatives)? Is it better to use only one type of test aiming at the best cost/effectiveness trade off, or a non-homogeneous testing policy? Naturally these questions need to be investigated at the interface of epidemiology, computer science, machine learning, mathematical modeling and statistics. As part of the work, the team will develop a model of transmission dynamics and control, tailored to COVID-19 in a way that accommodates diagnostic testing with varying fidelities and delays underlying a rapid testing regimen. The investigators will further integrate the resulting RTEM-SEIR model with EpiDMS and DataStorm for executing continuous coupled simulations.

This project is jointly funded through the Ecology and Evolution of Infectious Diseases program (Division of Environmental Biology) and the Civil, Mechanical and Manufacturing Innovation program (Engineering).

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2026860&HistoricalAwards=false

Project website:  https://rtem.live

Successfully tackling many urgent challenges in socio-economically critical domains (such as sustainability, public health, and biology) requires obtaining a deeper understanding of complex relationships and interactions among a diverse spectrum of entities in different contexts. In complex systems, (a) it is critical to discover how one object influences others within specific contexts, rather than seeking an overall measure of impact, and (b) the context-aware understanding of impact has the potential to transform the way people explore, search, and make decisions in complex systems. This project establishes the foundations of big data driven Context-Sensitive Impact Discovery (CSID) in complex systems and fills an important hole in big data driven decision making in many critical application domains, including epidemic preparedness, biological pathway analysis, climate, and resilient water/energy infrastructures. Thus, it enables applications and services with significant economic and health impact. The educational impacts of this project include the mentoring of graduate and undergraduate students, and the enhancement of graduate and undergraduate Computer Science curricula at both Arizona State University (ASU) and New Mexico State University (NMSU) through the incorporation of research challenges and outcomes into existing classes.

The technical goal of this project is to establish the theoretical, algorithmic, and computational foundations of big data driven context-sensitive impact discovery in complex systems. This project develops probabilistic and tensor-based models to capture context-sensitive impact from complex systems, often modeled as graphs, and designs efficient learning algorithms that can capture both the contexts and the impact scores among entities within these different contexts. The modeling of the context sensitive impact considers dynamic nature of relevant contexts and the diverse applications. This requires addressing several major challenges, including latent contexts of impact, heterogeneous networks of entities, dynamicity of impact in varying contexts, and high computational and I/O costs of context-sensitive impact discovery. Therefore, this project designs novel scalable probabilistic and tensor-based algorithms to capture and represent context-sensitive impact. These algorithms and the novel data platforms they are deployed in are efficient and scalable in terms of off-line and on-line running times and their space requirements. To achieve necessary scalabilities, the developed platforms employ novel multi-resolution data partitioning and resource allocation strategies and the research enables massive parallelism and efficient data access through new non-volatile memory based data management techniques.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1633381&HistoricalAwards=false

Natural disasters affect our society in profound ways. Between 2000 and 2009, disasters killed 1 million people, affected an additional 2.5 million individuals and caused a loss of about $1 trillion (2010 World Disasters Report). Effective disaster response requires a near-real-time effort to match available resources to shifting demands on a number of fronts. Experts today lack the means to provide emergency response agencies with validated strategies for disaster planning and response on a timely basis. Data-driven models and computer simulations for disaster preparedness and response can play a key role in predicting the evolution of disasters and effectively managing emergencies through a diverse set of intervention measures. This project will establish an approach that includes (a) planning disaster response, (b) public information and warning, (c) critical transportation services, (d) mass population care services, and (e) public health and medical services. Effective use of this integrated modeling approach may lead to enhanced safety, quality of life and community resilience. The project also provides an excellent context for doctoral, masters, and undergraduate level research and students will be introduced to career pathways through their participation in research, publication, and partnership with public agencies and data-driven science and engineering researchers.

This project will enhance disaster response and community resilience through multi-faceted research to create a big data system to support data-driven simulations with the necessary volume, velocity, and variety and integrate and optimize the key aspects and decisions in disaster management. This includes (a) a novel computational infrastructure capable of executing multiple coupled simulations synergistically, under a unified probabilistic model, (b) addressing computational challenges that arise from the need to acquire, integrate, model, analyze, index, and search, in a scalable manner, large volumes of multi-variate, multi-layer, multi-resolution, and interconnected and inter-dependent spatio-temporal data that arise from disaster simulations and real-world observations, (c) a new high performance data processing system to support continuous observation of the numerical results for simulations from different domains with diverse resource demands and time constraints. These models, algorithms, and systems will be integrated into a disaster data management cyber-infrastructure (DataStorm) that will enable innovative applications and generate broad impacts–through close collaborations with domain experts from transportation, public health, and emergency management–in disaster planning and response.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1610282&HistoricalAwards=false

GEARS is an enerGy-Efficient big-datA Research System at arizona state university, for studying heterogeneous and dynamic data by employing heterogeneous computing and storage resources and co-designing the software and hardware components of the system.

Big data technologies have been successfully applied to many disciplines for knowledge discovery and decision making, but the further growth and adoption of the big data paradigm face several critical challenges. First, it is challenging to meet the performance needs of modern big data problems which are inherently more difficult, e.g., learning of heterogeneous and imprecise data, and have more stringent performance requirements, e.g., real-time analysis of dynamic data. Second, power consumption is becoming a serious limiting factor to the further scaling of big data systems and the applications that it can support. These challenges demand a new type of big data systems that incorporate unconventional hardware capable of accelerating data processing and accesses while lowering the system’s power consumption. Therefore, this project is developing the needed computational infrastructure to support GEARS (an enerGy-Efficient big-datA Research System) for studying heterogeneous and dynamic data using heterogeneous computing and storage resources. GEARS is a one-of-kind, energy-efficient big-data research infrastructure based on cohesively co-designed software and hardware components. It enables a variety of important studies on heterogeneous and dynamic data and advances the scientific knowledge in computer science as well as other data-driven disciplines. It enhances the training of a large body of undergraduate and graduate students, including many from underrepresented groups, by supporting unique research and education activities. Finally, it also benefits the society by contributing new open-source solutions and with potential commercial applications in support of heterogeneous and dynamic data analysis.

The hardware of GEARS includes a cluster of data nodes equipped with heterogeneous processors and storage devices and fine-grained power management capability. The software is developed upon widely-used big data frameworks to support unified programming across CPUs, GPUs, and FPGAs and transparent data access across a deep storage hierarchy integrating DRAM, NVM, SSD, and HDD. GEARS also enables novel systems and algorithms research on learning heterogeneous and dynamic data, including (1) new algorithm partitioning and scheduling schemes for using heterogeneous accelerators and optimizing the performance and energy efficiency of big data tasks; (2) new I/O scheduling and data staging strategies for performance and energy efficiency of the deep big-data storage hierarchy; (3) multi-phase, out-of-core decomposition techniques for large-scale tensors; (4) real-time visual analytics system that links streaming media with simulations for anticipatory analytics; (5) multi-modal deep learning methods with heterogeneous social data; (6) new computational tools for real-time analysis of social unrest using social media; (7) scalable, adaptive, and interactive team detection and assemble system for designing high-performing teams using big network data; (8) rare category analysis and heterogeneous learning algorithms for fast and accurate rare event discoveries with large and heterogeneous social data; and (9) new distributed machine learning framework for learning semantic knowledge from Web-scale images/videos with incomplete/noisy textual annotations. All project results will be shared with the broader community via the project website (https://gears.asu.edu). Publications will be listed on the website with links to their publishers. Data and software downloads will listed on the website with instructions on how to use them. Source code will be hosted on GitHub and a direct link to the repository will also be listed on the project website.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1629888&HistoricalAwards=false

The building sector was responsible for nearly half of CO2 emissions in US in 2009. According to the US Energy Information Administration, buildings consume more energy than any other sector, with 48.7% of the overall energy consumption, and building energy consumption is projected to grow faster than the consumptions of industry and transportation sectors. As a response to this, by 2030 only 18% of the US building stock is expected to be relying on the current energy management technologies, with the rest either having been retrofitted or designed from the ground up using smart and cleaner energy technologies. These building energy management systems (BEMSs) need to integrate large volumes of data, including (a) continuously collected heating, ventilation, and air conditioning (HVAC) sensor and actuation data, (b) other sensory data, such as occupancy, humidity, lighting levels, air speed and quality, (c) architectural, mechanical, and building automation system configuration data for these buildings, (d) local whether and GIS data that provide contextual information, as well as (e) energy price, consumption, and cost data from electricity (such as smart grid) and gas utilities. In theory, these data can be leveraged from the initial design and/or retrofitting of buildings with data driven building optimization (including the evaluation of the building location, orientation, and alternative energy-saving strategies) to total cost of ownership (TCOs) simulation tools and day-to-day operation decisions. In practice, however, because of the size and complexity of the data, the varying spatial and temporal scales at which the key processes operate, (a) creating models to support such simulations, (b) executing simulations that involve 100s of inter-dependent parameters spanning multiple spatio-temporal frames, affected by complex dynamic processes operating at different resolutions, and (c) analyzing simulation results are extremely costly. The energy simulation data management system (e-SDMS) software will address challenges that arise from the need to model, index, search, visualize, and analyze, in a scalable manner, large volumes of multi-variate series resulting from observations and simulations. e-SDMS will, therefore, fill an important hole in data-driven building design and clean-energy (an area of national priority) and will enable applications and services with significant economic and environmental impact.

The key observations driving the research is that many data sets of urgent interest to energy simulations include the following: (a) voluminous, (b) heterogeneous, (c) multi-variate, (d) temporal, (e) inter-related (meaning that the parameters of interest are dependent on each other and constrained with the structure of the building), and (f) multi-resolution (meaning that simulations and observations cover days to months of data and may be considered at different granularities of space, time, and parameters). Moreover, generating an appropriate ensemble of simulations for decision making often requires multiple simulations, each with different parameters settings corresponding to slightly different, but plausible, scenarios. Therefore, significant savings in modeling and analysis can be obtained through data management software supporting modular re-use of existing simulation results in new settings, such as re-contextualization and modular recomposition (or “sketching”) of building models and if-then analysis of simulation traces under new parameters, new building floorplans, and new contexts. In developing the energy simulation data management system (e-SDMS), the research addresses the key data challenges that render data-driven energy simulations, today, difficult. This requires (a) a novel building models, simulation traces, and sensor/actuation traces (BSS) data model to accommodate energy simulation data and models, (b) feature analysis and indexing of sensory data and simulation traces along with the corresponding building models, and (c) algorithms for analysis and exploration of simulation traces and re-contextualization of models for new building plans and contextual metadata. This research will therefore, impact computational challenges that arise from the need to model, analyze, index, visualize, search, and recompose, in a scalable manner, large volumes of multi-variate series resulting from energy observations and simulations. E-SDMS consists of an (a) eViz server, which works as a frontend to e-SDMS, an (b) eDMS middleware for feature extraction, indexing, simulation analysis, and sketching, and an (c) eStore backend for data storage. To avoid waste and achieve scalabilities needed for managing large data sets, e-SDMS employs novel multi-resolution data partitioning and resource allocation strategies. The multi-resolution data encoding, partitioning, and analysis algorithms are efficiently computable, leverage massive parallelism, and result in high quality, compact data descriptions.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1339835&HistoricalAwards=false

The speed with which recent pandemics had immense global impact highlights the importance of realtime response and public health decision making, both at local and global levels. For instance, the SARS (Severe Acute Respiratory Syndrome) epidemic is estimated to have started in China in November 2002, had spread to 29 countries by August 2003, and generated a total of 916 confirmed deaths. A pandemic similar to the swine flu in 2009 is estimated to cost $360 billion in a mild scenario to the global economy and up to $4 trillion in an ultra scenario, within just the first year of the outbreak. Today, the key arsenal in the hands of decision makers who try to plan for and/or react to these outbreaks is software that enable model-driven epidemics and as well as the impacts of pharmaceutical and computer simulations for disease spreading. These software help predict geo-temporal evolution of non-pharmaceutical control measures and interventions, relying on data and models including social contact networks, local and global mobility patterns of individuals, transmission and recovery rates, and outbreak conditions. Unfortunately, because of the volume and complexity of the data and the models, the varying spatial and temporal scales at which the key transmission processes operate and relevant observations are made, today running and interpreting simulations to generate actionable plans are extremely difficult.

If effectively leveraged, models reflecting past outbreaks, existing simulation traces obtained from simulation runs, and real-time observations incoming during an outbreak can be collectively used for obtaining a better understanding of the epidemic’s characteristics and the underlying diffusion processes, forming and revising models, and performing exploratory, if-then type of hypothetical analyses of epidemic scenarios. More specifically, the proposed epidemic simulation data management system (epiDMS) will address computational challenges that arise from the need to acquire, model, analyze, index, visualize, search, and recompose, in a scalable manner, large volumes of data that arise from observations and simulations during a disease outbreak. Consequently, epiDMS fill an important hole in data-driven decision making during health-care emergencies and, thus, will enable applications and services with significant economic and health impact.

The key observation is that the modeling and execution can be significantly reduced using a data-driven approach that supports data and simulation reuse in new settings and contexts. Relying on this observation, in order to support data-driven modeling and execution of epidemic spread simulations, this team will develop

+ an epidemic data and model store (epiStore) to support acquisition and integration of relevant data and models.

+ a novel networks-of-traces (NT) data model to accommodate multi-resolution, interconnected and inter-dependent, incomplete/imprecise, multi-layer (networks), and temporal (time series or traces) epidemic data.

+ algorithms and data structures to support indexing of networks-of-traces (NT) data sets, including extraction of salient multi-variate temporal features from inter-dependent parameters, spanning multiple simulation layers and geo-spatial frames, driven by complex dynamic processes operating at different resolutions.

+ algorithms to support the analysis of networks-of-traces (NT) datasets, including identification of unknown dependencies across the

input parameters and output variables spanning the different layers of the observation and simulation data.

The proposed NT data model and algorithms will be brought together in an epidemic simulation data management system (epiDMS). For broadest impact, the proposed epidemic simulation data management system (epiDMS) will be designed in a way that interfaces with the popular Global Epidemic and Mobility (GLEaM) simulation engine, a publicly available software suit to explore epidemic spreading scenarios at the global scale. To achieve necessary scalabilities, epiDMS will employ novel multiresolution data partitioning and resource allocation strategies and will leverage massive parallelism.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1318788&HistoricalAwards=false

Aviation has for decades been an early adopter of intelligent technologies, just to mention few such as the coupled autopilots, full authority digital engine control, adaptive autopilots, etc. History of modern intelligent onboard systems dates back to 2003 when NASA began flight testing using an artificial neural network that analysed aircraft flight characteristics and could create alternative flight control laws. Research into intelligent onboard technologies has intensified over past years. This project has for its objectives to (a) use real flight profile data to assess quality and completeness of provided flight profile data in terms of their suitability for onboard data analytics for Class E and Class D operations and (b) use fuel saving use case (a Class E operation) as a practical example to demonstrate potentials of on board machine learning.

A recommender system helps users to identify useful and interesting items from a considerably large search space. Recommender systems have been widely used in various commercial services. A recommender system exploits the users’ opinions in order to extract a set of interesting items for each user. This project conducts research, develop requisite knowledge and build software infrastructure to support efficient, salable, and usable data management for personalized recommendation applications. Recommender systems have already been widely used with a strong broad impact on all web users and the project aims to take personalized recommendation applications recommender systems to its next stage and widening its scope to new applications. The project enhances the research infrastructure by distributing a free and portable software artifact. All proposed ideas will be realized inside an open-source recommendation-aware database system maintained at Arizona State University. It is envisioned that the proposed system will be used by researchers world wide as a vehicle for evaluating their research and exchanging new modules related to recommender systems. It is also envisioned that several commercial database systems will adopt the ideas from this project. The project will have a significant educational component. Researchers in both data management and recommender systems will be trained through the proposed project, through curricular innovations as well as workshops and tutorials. Students will be introduced to career pathways through their participations in research.

The project tackles the following system challenges to support recommendation applications: (1) Flexibility and Usability: The user should be able to declaratively define a variety of recommenders using popular recommendation algorithms that fit the application needs. The system must be able to integrate the recommendation functionality with other data attributes/sources as well as performing the recommendation functionality and other data access operations side by side. (2) Efficiency and scalability: The system is expected to produce personalized recommendations to a high number of users concurrently over a large pool of items. Unfortunately, recommender models are not easily updatable, and hence they are rebuilt periodically. As a result, the model loses its accuracy over time till the next rebuild process. This is not acceptable in modern applications (e.g., social media) where new items and ratings are streaming into the system. To tackle these challenges, the project injects the recommendation functionality inside the core functionality of a database system by: (a) indexing the set of recommenders to efficiently answer of ad-hoc recommendation queries, (b) encapsulating the recommendation functionality inside a pipeline-able query operator that integrates well with other database operators, and designing query optimization techniques that include the recommendation functionality. Moreover, since a common operation to train recommendation models is to factorize multi-relational user, item, and attribute data, in the forms of tensors, this proposal develops a scalable parallelizable data processing software framework that provides co-optimization of tensor-algebraic and relational algebraic operations. The project also leverages database systems to support context (e.g., spatial location and social network)-aware recommendations.

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1654861&HistoricalAwards=false

This Partnerships for Innovation: Building Innovation Capacity (PFI:BIC) project from Arizona State University focuses on building a platform that will integrate data from multiple sources and explore data analysis techniques that can more accurately detect indications of financial fraud. According to Federal Trade Commission’s (FTC) annual “Consumer Sentinel Network Data Book”, the most comprehensive database of U.S. fraud trends, American consumers submitted more than 1.5 million complaints – a 62 percent increase in just three years, and they reported losing over $1.6 billion to fraud in 2013. Detecting increasingly complex fraud schemes requires services that are able to integrate and enrich data from disparate financial and other data sources and hunt for recurring and often interconnected anomalous patterns in large networks. The proposed platform will enable integration and enrichment of limited private financial data with larger publicly available data sets to detect fraud and reduce losses due to fraudulent transactions. The project will also include training and research experience for undergraduate and graduate students.

The data linkage and financial pattern discovery platform which is to be developed via visual analytics will enable “smart” fraud detection and prevention services. Today, in order to obtain a single unified view of fraud activity across the enterprise and manage fraud on a cross-institution basis, fraud detection companies collect, verify, and analyze consumer data and financial information. Researchers recognize, however, that new insights into fraud and risk patterns require the ability to integrate financial data with domain independent data through real-time entity/identity discovery, resolution, cross-linking and schema mapping techniques. Therefore, the importance of the research discovery underpinning this project includes solving platform and processing challenges that arise from the need to integrate, filter, analyze, and visualize, in a secure and scalable manner, large private knowledge networks, also incorporating uncontrolled, unrestricted, untrusted, unstructured and unpredictable data from external domains. The ability to treat together financial and domain independent data will lead to enriched unified data, unprecedented predictive accuracy in fraud prevention and detection, and an entirely new suite of risk management services and products.

The partners at the inception of the project are Arizona State University (ASU) (School of Computing, Informatics, and Decision Systems Engineering and, notably, the investigators also are members of ASU’s Information Assurance Center, certified by NSA and DHS); and a small business, Early Warning Services, LLC (Scottsdale, AZ).

Grant info: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1430144&HistoricalAwards=false