Raghu Machiraju - US grants

CARTER 9730381 This award provides funding for an Engineering Research Centers (ERC) Program/Centers for Research Excellence in Science and Technology (CREST) Program Partnership between the Engineering Research Center for Computational Field Simulation at Mississippi State University (MSU) and Florida A&M University and Florida International University CREST Center for Distributed Computing, entitled "An ERC-CREST Partnership in Distributed and Computational Systems." This partnership will be initially founded on collaboration in four research projects: (1) The Hector Run-Time Environment; (2) Data Compression and Visualization using Wavelets; (3) Engineering Distributed and Parallel Software; and (4) Application of Parallel Computing Environments to Molecular Reaction Theory. The partnership will increase interactions and improve the educational experiences of students at all three institutions with the additional exposure serving to attract more students, particularly minority students into research careers.

The field of visual computing is concerned with the storage, analysis, manipulation of sampled definitions of 3D objects and the resulting image synthesis. Given the ubiquitous use of imagery for graphics in architecture, engineering design and manufacturing, and entertainment, there is an ever-growing need for realism in displayed images. For serious uses of imagery in engineering and medicine, there is an additional need for assurances for accuracy and for providing ways to validate the image. This project will increase the level of understanding about the foundational aspects of image synthesis, an essential component of visual computing. Volume rendering and visualization algorithms are being increasingly used for image synthesis in visual computing systems. The project plans to specifically address the following issues: * Accuracy in volume rendering algorithms-Lack of concerted efforts to define and measure image quality has stymied the widespread use of visualization techniques and visual computing. * Volume Analysis- Given the noisy nature of the datasets and large sizes, there is a need to seek minimal correlated representations of the volumes and even rendering operators. * Perceptual metrics - The displayed image should preset meaningful information to the end-user. It would be useful to employ objective numerical tools to guide the generation of new images if desired. One approach to achieving this goals will be to use multiresolution methods including the wavelet transform. The work will be initially targeted towards the requirements of computational fluid dynamics engineers, marine geologists and medical imaging experts, given the access of Principal Investigator (PI) to the ERC, NASA Stennis and the University of Mississippi Medical center (UMMC) respectively.

9818489
Skjellum, Anthony
Hodges, Julia
Mississippi State University

A Gigabit/s, VIA-Enabled Cluster Architecture for Research in High
Performance Systems Software, Scalable Knowledge Discovery, Visualization,
and Parallel Planning under Uncertainty

This research instrumentation enables research projects in:

- Multi-Grain Parallel Processing,
- Scalable Knowledge Discovery,
- Parallel Volume Visualization for Computational Field Simulation, and
- Parallel Processing for Markov Model Planning.

Concurrency is an important enabling technology for several areas of computational science, unlocking the potential for new science. This research instrumentation proposal brings together three innovative
computational activities together with a fourth research activity that enhances the capability and understanding of parallel processing environments, notations, and services. Six investigators at Mississippi
State University, Department of Computer Science, undertake the following projects: enhanced infrastructure for multi-grain parallel processing (including extensions to MPI, MPI-2, and MPI/RT), scalable knowledge
discovery (artificial intelligence plus high performance computing to achieve automatic document classification and unsupervised learning algorithms), parallel volume visualization for computational field
simulation, and parallel processing for Markov Model Planning (study of fast algorithms for exact solutions to planning under uncertainty). To achieve this new science, a parallel processing cluster, consisting of thirty, two-way SMP Pentium II systems is to be assembled, together with multiple high speed networks, one of which supports the Virtual Interface Architecture for low processor overhead and high bandwidth. Commercial-grade message passing software (MPI) and scheduling software are used to provide a production environment under Windows NT immediately, in complement to the research experiments involving parallel processing that further enable the environment over time, including part-time Linux-based operation.

Large-scale computational simulations of physical phenomena produce data of unprecedented size (terabytes and petabytes). Unfortunately, development of appropriate data management and visualization techniques has not kept pace with the size and complexity of such datasets. For many simulations, the storage of the data itself, not to mention seeing and understanding it, is a serious problem. This project responds to this by developing a prototype, integrated system (EVITA) to facilitate exploration of terascale datasets. This system will be concerned with the storage and representation of very large datasets, their analysis and manipulation and rendering and presentation of the resulting images. This will allow unprecedented access to our CFD data sets, which we expect to lead to new fundamental understanding.

The project focuses on time-varying datasets from computational fluid dynamic/hydrodynamic and oceanographic applications on structured, rectilinear or curvilinear grids; however, the EVITA system is applicable as a general visualization environment for other terascale datasets with similar underlying structure. Additionally, the system is amenable to distributed- or even remote-visualization uses. The cornerstone of the EVITA system is a representational scheme that allows ranked access to macroscopic features in the dataset. The data and grid are transformed using wavelet techniques while a feature-detection algorithm is used to identify and rank contextually significant features directly. This ranking is used to generate a four-dimensional significance map that incorporates application-specific knowledge, which in turn allows an encoding with efficient compression and a progressive representation of significant features in the data. From this, the EVITA system creates as a preview an efficient, "lossy" image of the desired data and lets the user select regions of interest (ROls) for further enrichment. These ' Rols are given higher priority in the encoded bitstream, and subsequently displayed faster. Preliminary results demonstrate the efficacy of this approach. The project takes a novel approach to terascale visualization by capitalizing on the expertise of a multidisciplinary team to integrate research from several disciplines into the EVITA system. This both brings many perspectives to bear on the problem, and ensures that our results are practically useful.

Despite many advances in the visualization community over the past decade,
effective and efficient representations for three-dimensional flow fields are still elusive.
Many techniques show promise, but do not represent a complete picture of
the flow.
This project consists of a systematic investigation of the current
techniques, their advantages and disadvantages, research and analysis of combined techniques which complement each other, and research into new techniques.
Cognitive analysis and user studies will be undertaken to investigate existing research. Many of these techniques, including those of the PIs, proposed a method of representing or rendering a flow
field, with little selfexamination of the limitations towards garnering a true understanding of the flow
field.
Several of these techniques have obvious limitations, but present an improvement over
the dearth of any good techniques for three-dimensional flow visualization. By
combining several techniques appropriately, more improvements can be
made.
Are these improvements sufficient? What combinations work and which ones do not?
These questions will be investigated and quantified in this research.
Finally, we will ask: How can new techniques be developed that will overcome some of these
limitations?Our subjective understanding of the field allows us to propose several
avenues of investigation. These include an-isotropic volume renderings, embedded
with flow differentiators; schematic texture algorithms for opaque and semi-transparent stream
surfaces; level-of-detail models for representing the flow; multi-resolution representation
of underlying features; and new representations that provide a more meaningful context
for the flow.
This research will greatly aid computational scientists across many disciplines: computational fluid dynamics scientists, environment scientists, atmospheric scientists, computational geological scientists and biomedical researchers.
A primary component of their research is the investigation or simulation of natural phenomena.
This phenomena is generally dynamic, leading to complex and interesting flow fields. For
three-dimensional computations, effective tools to understand, represent and convey these
flows are lacking. This research will provide a firm foundation for research in this area, as well as lead to new algorithms offering a much clearer representation for flows.

Numerical simulations are replacing traditional experiments in gaining insights into complex physical phenomena. Given recent advances in computer hardware and numerical methods, it is now possible to simulate physical phenomena at very fine temporal and spatial resolutions. As a result, the the amount of data generated is overwhelming.

Scientists are interested in analyzing and visualizing the data produced by such simulations to better understand the process that is being simulated. Analyzing such large scale data is hard. Not only the methods used are computationally expense, current programming tools make the analysis difficult to specify and modify. Thus, there is a dire need for a systematic approach, along with supporting algorithms and methodologies for flexible parallel implementations, to achieve scalable and interactive analysis on large scientific datasets.

In this project, we propose the construction of such a scalable toolkit, namely the Computational Analysis Toolkit (CAT). This toolkit proposes to exploit ongoing work in feature analysis, scalable data mining and parallel programing environments. The crux of the approach is feature-mining; a process where by regions are delineated through various stages of detection, verification,
de-noising, and tracking of points of interest. Additionally, we propose the use of some key data mining mining algorithms for achieving enhanced and robust implementations of feature-mining algorithms.

It is our objective that the CAT toolkit should not only allow for the detection of features but also provide for a means to control the analysis in an interactive setting. For example, demographic and lifetime analysis of certain critical features as determined by the user/scientist may be an important way of understanding the underlying process being simulated. These critical features, once tagged via a suitable interface, can be profiled and a concise representation this profile can then be presented to the user as needed.

We believe that for long-term use of a tool for feature and data mining, it is important that a) the algorithms are parallelized on a variety of platforms, b) the parallel implementations are easy to maintain and modify, and c) APIs are available for users to rapidly create scalable implementations of new mining algorithms. We are proposing to achieve these goals by using and extending a parallelization framework developed locally. This framework, referred to as FRamework for Rapid Implementations of Datamining Engines (FREERIDE), offers high-level APIs and runtime techniques to enable parallelization of algorithms for data mining and related tasks. It allows parallelization on both distributed memory and shared memory configurations, and further supports efficient processing of disk-resident datasets.

The proposal, besides providing a useful toolkit, is likely engender the use of methodologies for large data exploration. Our efforts are likely to contribute to literature in scalable data and feature mining algorithms, and feature profile summarization.

In science the challenge is always finding a signal in the noise. Examples include hurricane forecasting and monitoring both intelligence and seismic activity.
Our proposal addresses these issues through a broad framework we call generalized feature mining. The framework has two major components: feature mining, and shape-based data mining and analysis. At its core, feature mining detects features for a specific application domain. Each instance involves a specific extended shape description tailored to it. For evolutionary simulations, feature mining also tracks features across multiple temporal scales. Shape-based data mining and analysis learn from the process. The aim is to correlate information from the extended shape descriptors with transient detection to find or refine spatio-temporal rules for the evolution of features. Environmental influences, such as walls, must be built into the rules so they are predictive.
To close the loop, the detected features can be displayed as they are found or refined. The evolutionary rules predicted by our framework can lead to new science { not only understanding the underlying phenomena but also leading to computationally simpler models that encapsulate the essentials.

ABSTRACT

This proposal seeks to advance security surveillance monitoring by introducing event-based reasoning. The team will use a formal event-discovery protocol to uncover event categories and the temporal structure of events. This results in an event template hierarchy. The event template hierarchy is supported by the enabling technologies of smart sensors, a reconfigurable network, and the use of persistent models for tracking. The result is an autonomous sensor network that can be effectively coupled to human operators in order to allow top-down control of the resources as well as the ability to modify the models for event and background activities. While the methodology is suitable for a wide variety of application domains, the work is grounded in a campus security and surveillance paradigm.

By integrating research from Cognitive Science, Geography, and Computer Science (Graphics, and Vision), the team can create a paradigmatic shift in the way that surveillance systems are viewed and developed. The data stream is no longer composed merely of video and perhaps some low-level alarms; the focus is now extended to include events. Data and information no longer move toward a usersitting in front of a wall of monitors. Event contexts, set by higher-level events as well as by operators-in-the-loop, direct and focus attention in order to detect differences from a dynamic model of background activity. The result is that the information is more meaningful, the surveillance systems more focused, and the cognitive skills of the operators more efficiently utilized. A prototype system will be made available for pertinent security personnel to train and test. The work will contribute to training methodologies of security personnel. Under the purview of broader impact, the proposed work strives to include under-represented and minority student groups through targeted training in the use of video technology. Finally if successful, event-based strategic surveillance networks can provide alternatives to racial profiling. The individuals are judged only by their actions as encoded in the event models.

Several application domains require processing of tera/peta-bytes of data. Developing data-intensive applications with such massive datasets poses several challenges, with respect to data management, processing, and resource allocation. Furthermore, algorithms and their pertinent parameters that yield robust and near-optimal results are often determined through an iterative process for which interactive response times are essential. Thus, there is a need for being able to rapidly create scalable and parallel implementations of a variety of data analysis algorithms.

An emerging and critical area requiring large-scale data analysis is medical imaging. Complex algorithms and novel tools are required to be able to analyze such data. In the project, we consider data obtained from fMRI (functional MRI). Driven by this domain, this project focuses on how algorithm design, API design, and runtime system development can be combined to provide effective data-intensive solutions for spatio-temporal data analysis. Particularly, we target the following four questions:
1) How can we exploit the map-reduce paradigm to accelerate advances in neuroimaging and related medical fields? .
2) What are some of the challenges in using map-reduce paradigm for neuroimage analysis?
3) What alternative interface to the current map-reduce API can provide still provide ease of expression of data analysis algorithms, while enabling better performance?
4) Can we use the map-reduce and similar paradigms starting from high-level languages, such as Matlab?


This project will also make substantial contributions towards teaching, human resource development, and increasing diversity. Most of the requested funds will be used for supporting Ph.D students on this project. PI Agrawal expects to involve at least one of his three current female Ph.D students in this project. PI Machiraju also engages with several undergraduate students at the medical campus of OSU.

Abstract - Machiraju, Rangarajan, and Thompson

As computer power continues to increase, the complexity of simulations also increases thereby producing datasets of unprecedented size. Without effective analysis tools, results from these large-scale simulations cannot be utilized to their fullest extent. This research addresses the problem of large-data visualization and exploration by employing interactive multi-scale machine learning, which exploits an efficient feature-based, multi-resolution representation of the data. The investigators are leveraging methods from the field of machine learning to perform two distinct tasks: identify regions of interest and enhance robustness of feature detection algorithms. The primary outcome of this effort is the realization of a framework for exploring large datasets. Further, this work is introducing a large body of work in machine learning to the field of visualization. Successful completion of this research will help overcome the brittleness of existing visualization methods and foster expedient discovery in many areas of science and engineering.

The multi-resolution techniques developed here will employ a two-fold strategy. First, semi-supervised learning based on training with the domain expert is used to develop strategies for selective spatial and temporal refinement of the data. A classifier is constructed to tag the output of the coarse resolution feature detection (i.e. regions) as either interesting or not interesting. Then at the finest scale, interesting local data chunks containing features of interest are identified for further analysis. Second, several local feature detection algorithms, or weak classifiers, are combined into a single, more robust compound classifier using adaptive boosting, or AdaBoost, and a data adaptive variant called CAVIAR that facilitates validated feature detection. Ideally, the compound classifier combines the best of all weak classifiers as they respond to the underlying physical signal. This research is demonstrating the effectiveness of these methods by applying existing local detection algorithms for visualizing vortices in turbulent flow fields.

Proteins are the molecular machines that are responsible for a vast array of functions that are necessary for life. Understanding how they work is critical to both a better scientific understanding of the fundamental processes of life, and to modifying or improving their function. Despite the fact that proteins are physically 3-dimensional structures of cooperating parts, the current state of the art for representing and studying proteins uses a description that is simply a sequential list of the parts used in their assembly. This sequential-list style of description has biased the development of tools for protein analysis to accentuate the sequential properties of these molecules, and to ignore the fact that the parts must work together in unison for the protein to function. This project will adapt a recently-developed statistical technique, the Conditional Random Field (CRF), that can quantitatively represent densely-connected networks of features, and a recently-developed visualization tool that enables interactive exploration of these networks, for the task of describing proteins. Structurally, Conditional Random Fields appear to recapitulate the process by which evolution has selected for parts that cooperate in proteins, and protein descriptions based on CRFs will be able to predict whether a change to a protein - a mutation - would have been tolerated by evolution, or selected against as non-functional. This information will aid in predicting the effect of a mutation, or multiple mutations to a protein, using much more of the available information, than is currently utilized by state-of-the-art tools.

This work will broadly impact the study of proteins, improving a range of activities from basic scientific studies of function, to endeavors in protein engineering. In addition, the "change in protein sequence to change in protein function" problem is a "model organism" for many other types of biological and non-biological systems where rich interactions between parts of the system demand a sophisticated statistical approach. To-date, in most of these fields, models that are similarly limited to those currently used in proteins are the de-facto standard. Developing the tools necessary for applying CRFs to protein data, and methods of establishing testable ground-truth in this system, will enhance the application of CRFs to many other domains where they may provide a significant advantage over current methods. The products of this project will be made freely available to the research community as online tools, and the methods will be incorporated in coursework, first in the Biophysics Graduate Program at The Ohio State University, and as the teachable component matures, made available as lesson-plan material appropriate for both primary and secondary education. By developing a tool that makes interdependencies between features visually explorable and modifications of these dependencies quantifiably predictable, we will promote more thorough consideration of the true complexity of data and systems in many domains.

The biological, physical and socials sciences are being overwhelmed by the large amounts of data: i) from new generations of instruments and sensors, with rapidly decreasing costs and rapidly increasing resolutions; ii) the creation of large scale instrumented environments; and iii) the use of high performance simulation that produces large datasets. Data Science is an emerging field that is developing to meet this challenge, which integrates domain knowledge from the relevant disciplines with statistics/mathematics and computer science/informatics. Translational data science is a new term that is being used for an emerging field that applies new data science principles, techniques and technologies to challenging scientific problems that hold the promise of having an important impact on human or societal welfare. The term is also used when data science principles, techniques and technologies are applied to problems in different domains in general, including,but not restricted to,science and engineering research. The team will hold the first Workshop on Translational Data Science (TDS 17) in Chicago on June 26-27, 2017 as a first, important step to foster the creation of the discipline of translational data science. This workshop will bring together about 50 scientists from academia, industry, foundations, and federal funding agencies to discuss important issues, challenges and opportunities, including the scope of translational data science. A key outcome from the workshop will be a white paper about translational data science. Another key outcome will be planning for future workshops around translational data science and the beginning of the creation of Translational Data Science Community.

Franklin County, Ohio, home of the state's capital at Columbus, has one of the highest infant mortality rates in the country at 8.3 deaths per 1,000 live births. As outlined in a recent article in the Journal of the American Medical Association (2016), the U.S. lags behind in many important measures of population health, including infant mortality, despite the fact that one-fifth of our dollars are spent on healthcare. While the data needed to address these important public health problems are available, to date we have not seen adequate investment in informatics approaches to combine and analyze these multilevel (e.g., individual lifestyle factors, neighborhood characteristics) data. The current planning project represents an effort to plan (with an intent to implement) just such an approach to address the high rate of infant mortality in Greater Columbus and nationwide. As the recent winner of the Department of Transportation's Smart Cities Challenge, Columbus already has a well-developed plan to employ technology to improve local transportation options, which is anticipated to ameliorate some contributing factors such as lack of access prenatal and other health care. The financial support from this and follow-up grants will permit us to expand that effort and leverage technological advances to further identify and design interventions to address risk factors for poor maternal and infant health outcomes. Further, by including students and other trainees in our work, we will be training the next generation of scientists to develop innovative solutions to address complex societal problems.

The objectives of this one-year planning project are to: 1) identify data gaps that present local barriers to achieving optimal maternal and infant health and their effect on proximate causes of infant mortality in the community; 2) align with key stakeholders and partners in the community with a goal to identify opportunities where technology, especially connectivity and mobility, could be leveraged to address barriers and speed-up progress; and 3) utilize our technological and content expertise to design and implement novel interventions for improving maternal and infant health. Since 2014 there have been a number of local efforts to address Franklin County's high rate of infant mortality to no avail. As such, there is a dire need for a coordinated novel multidisciplinary approach. The primary focus and intellectual contribution of this planning grant is to work closely with stakeholders to plan strategies to identify the key contributors of infant mortality in Franklin County (i.e., likely specific social determinants of health) and to develop novel interventions driven by innovations in BIGDATA technology.

The recent explosion in worldwide data together with the end of Moore's Law and the near-term limits of silicon-based data storage being reached are driving an urgent need for alternative forms of computing and data storage/retrieval platforms. In particular, exabyte-scale datasets are increasingly being generated by the biological sciences and engineering disciplines including genomics, transcriptomics, proteomics, metabolomics, and high-resolution imaging, as well as disparate other scientific fields including climate science, ecology, astronomy, oceanography, sociology, and meteorology, amongst others. In this data revolution, the continuously increasing size of these datasets requires a concomitant increase in available computational power to store, process, and harness them, which is driving a need for revolutionary new, alternative substrates for, and forms of, computing and data storage. Unlike traditional data storage and computing materials such as silicon, the human brain offers a remarkable ability to sense, store, retrieve, and compute information in a manner that is unrivaled by any human-made material. In this research project, analogous modes of information sensing, data storage, retrieval, and computation will be explored in non-traditional computing molecular systems and materials. The over-arching goal of the research is to discover revolutionary new modes of data storage/retrieval, sensing, and computation that rival conventional silicon-based technology, for deployment to benefit society broadly across all domains of data science. Graduate students and postdocs across five institutions will be trained and mentored in a highly interdisciplinary manner to attain this goal and prepare the next-generation of data scientists, chemists, physicists, and engineers to harness the ongoing data revolution. The research will be disseminated to a broad community through news outlets and integration of high school student internships in participating research laboratories.

Large-scale datasets from spatial-temporal calcium imaging of the mouse brain will be recorded into DNA-based, nanoparticle-based, and phononic 2D and 3D soft and hard materials. Continuous spatial-temporal data will first be transformed into discrete data for mapping onto DNA-conjugated fluorophore networks, dynamic barcoded nanoparticle networks, and phononic 2D and 3D materials. Sensing, computation, and data storage/retrieval will be demonstrated as proofs-of-principle in exploiting the chemical properties of molecular networks and materials to recover the encoded neuronal datasets and their sensing and computing processes. Success with any of these three prototypical materials would revolutionize the ability to encode arbitrarily complex, large-scale datasets into complex molecular systems, with the potential to scale across diverse data domains and materials frameworks. The investigators' Autonomous Computing Materials framework will thereby enable the encoding of arbitrary "big data" sets into diverse materials for data storage, sensing, and computing. This project maximizes opportunities for disruptive new computing and data science concepts to emerge from a multi-disciplinary, collaborative team spanning data science, neuroscience, materials science, chemistry, physics, and biological engineering.

This project is part of the National Science Foundation's Harnessing the Data Revolution (HDR) Big Idea activity, and is jointly supported by HDR and the Division of Chemistry within the NSF Directorate of Mathematical and Physical Sciences.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

The analysis of high-resolution images in both two and three-dimensions is becoming important for many scientific areas, such as in medicine, astronomy and engineering. Discoveries in these disciplines often require analyzing millions of images. The analysis of these images is complex and requires many steps on powerful computers. Some of these steps require looking through lots of images while some of these steps require deep analysis of each image. In many cases, these analyses have to be completed quickly, i.e. in "real-time", so that information and insights can be provided to humans as they do their work. These kinds of operations require powerful computers consisting of many different, heterogeneous but simple computing components. These components need to be configured and reconfigured so that they can efficiently work together to do these large-scale analyses. In addition, the software that controls these computers also has to be intelligently designed so that these analyses can be run on the right types of configurations. This project aims to acquire the necessary computing components and assemble such a powerful computer (named RADiCAL). Research done using RADiCAL will result in important scientific discoveries that will make us more prosperous, improve our health, and enable us to better understand the world and universe around us. Doing this research will also educate many students, including those from under-represented groups, who will become part of a highly-trained workforce capable of addressing our nation's needs long into the future.

The intellectual merit of RADiCAL is in the design a novel, high-performance, next-generation, heterogeneous, reconfigurable hardware and software stack to provide real-time interaction, analytics, machine/deep learning (ML/DL) and computing support for disciplines that involve massive observational and/or simulation data. RADiCAL will be built from commodity hardware, and designed for reconfiguration and observability. RADiCAL will enable a comprehensive research agenda on software that will facilitate rapid and flexible construction of analytics workflows and their scalable execution. Specific software research include: 1) a library with support for storage and retrieval of multi-resolution, multi-dimensional datasets, 2) scalable learning and inference modules, 3) data analytics middleware systems, and 4) context-sensitive human-in-the-loop ML models and libraries that encode domain expertise, coupling tightly with both lower level layers and the hardware components to facilitate scalable analysis and explainability. With the proposed hardware acquisition and software research, the transformative goal will be to facilitate decision-making and discovery in Computational Fluid Dynamics (CFD) and medicine (pathology). With respect to broader impacts, RADiCAL will provide a unique research, testing, and training infrastructure that will catalyze research in multiple disciplines as well as facilitate convergent research across disciplines. The advanced imaging applications and techniques for expert-assisted image analysis will be broadly applicable to other human-in-the-loop systems and have the potential to advance medicine and health. Projects that use RADiCAL will also provide unique test-beds for valuable empirical research on human-computer interaction and software engineering best practices. Well-established initiatives at The Ohio State University will facilitate the recruitment of graduate and undergraduate students from underrepresented groups for involvement in using the cyberinfrastructure. The heterogeneous and reconfigurable research instrument will be utilized to create sophisticated educational modules on how to co-design computational science experiments from the science goals to the underlying cyberinfrastructure. Tutorials and workshops will be organized at PEARC, Supercomputing and other conferences to share the research results and experience with the community.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Artificial Intelligence (AI) is used in many aspects of modern life such as language translation and image analysis. In addition to consumer and business applications, researchers are increasingly using AI techniques in their scientific processes. The growth in AI is heavily dependent on new Deep Learning (DL) and Machine Learning (ML) schemes. As datasets and DL and ML models become more complex the computing requirements for AI increase and researchers turn to high performance computing (HPC) facilities to meet these needs. This is leading to a critical need for a Cyberinfrastructure (CI) workforce that supports HPC systems with expertise in AI techniques and underlying technology. This project will pilot an AI bootcamp for CI professionals that is targeted based on the professional's job requirements. After attending the bootcamp CI professionals will be better equipped to provide computing and data services to AI research users. This in turn will broaden adoption and effective use of advanced CI by researchers in a wide range of disciplines and will have an impact on science and corresponding benefits to society from their successes. The training materials developed during this project will be openly shared with the CI community so that others can use and adapt the materials for similar training activities.

This project is novel in taking a holistic approach to addressing the AI expertise gap for CI professionals. The project will develop an AI Bootcamp for CI professionals with the overarching goal of increasing the confidence and effectiveness of their support of AI researchers. The project leverages the CI professionalization efforts of the Campus Research Computing Consortium (CaRCC) to organize the training outcomes based on four "facings" (Strategy/Policy facing, Researcher facing, Software/Data facing, and Systems facing). The project will identify learning outcomes for each CI facing and organize training tracks customized to specific roles. For this pilot the project is focused on developing a comprehensive training experience for Software/Data facing CI professionals. The AI Bootcamp will be offered virtually over twelve weeks. The instructional materials will be shared openly as notebooks, slide-decks and containers as appropriate so that they can be used for other training offerings. The project team is comprised of CI professionals, experienced in training CI users and providing CI operations, and Computer Science faculty members, experienced in offering courses in Data Analytics, AI and High Performance AI with active AI-based research programs. Drawing on extensive experience and materials in hands-on experiential learning for AI, the project team will create a comprehensive curriculum spanning foundational AI, software frameworks, and high performance computing for AI in a modularized virtual format to minimize barriers to access for the CI professional learner.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Although the world is witness to the tremendous successes of Artificial Intelligence (AI) technologies in some domains, many domains have yet to reap the benefits of AI due to the lack of easily usable AI infrastructure. The NSF AI Institute for Intelligent Cyberinfrastructure with Computational Learning in the Environment (ICICLE) will develop intelligent cyberinfrastructure with transparent and high-performance execution on diverse and heterogeneous environments. It will advance plug-and-play AI that is easy to use by scientists across a wide range of domains, promoting the democratization of AI. ICICLE brings together a multidisciplinary team of scientists and engineers, led by The Ohio State University in partnership with Case Western Reserve University, IC-FOODS, Indiana University, Iowa State University, Ohio Supercomputer Center, Rensselaer Polytechnic Institute, San Diego Supercomputer Center, Texas Advanced Computing Center, University of Utah, University of California-Davis, University of California-San Diego, University of Delaware, and University of Wisconsin-Madison. Initially, complex societal challenges in three use-inspired scientific domains will drive ICICLE’s research and workforce development agenda: Smart Foodsheds, Precision Agriculture, and Animal Ecology.

ICICLE’s research and development includes: (i) Empowering plug-and-play AI by advancing five foundational areas: knowledge graphs, model commons, adaptive AI, federated learning, and conversational AI. (ii) Providing a robust cyberinfrastructure capable of propelling AI-driven science (CI4AI), solving the challenges arising from heterogeneity in applications, software, and hardware, and disseminating the CI4AI innovations to use-inspired science domains. (iii) Creating new AI techniques for the adaptation/optimization of various CI components (AI4CI), enabling a virtuous cycle to advance both AI and CI. (iv) Developing novel techniques to address cross-cutting issues including privacy, accountability, and data integrity for CI and AI; and (v) Providing a geographically distributed and heterogeneous system consisting of software, data, and applications, orchestrated by a common application programming interface and execution middleware. ICICLE’s advanced and integrated edge, cloud, and high-performance computing hardware and software CI components simplify the use of AI, making it easier to address new areas of inquiry. In this way, ICICLE focuses on research in AI, innovation through AI, and accelerates the application of AI. ICICLE is building a diverse STEM workforce through innovative approaches to education, training, and broadening participation in computing that ensure sustained measurable outcomes and impact on a national scale, along the pipeline from middle/high school students to practitioners. As a nexus of collaboration, ICICLE promotes technology transfer to industry and other stakeholders, as well as data sharing and coordination across other National Science Foundation AI Institutes and Federal agencies. As a national resource for research, development, technology transfer, workforce development, and education, ICICLE is creating a widely usable, smarter, more robust and diverse, resilient, and effective CI4AI and AI4CI ecosystem.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.