Presentation

Integration of Geospatial Data: Examples and Implications

Wendy Martinez, Bureau of Labor Statistics and President-Elect ASA

Abstract

The Federal Committee on Statistical Methodology (FCSM) is charged with informing and advising the Office of Management and Budget and the US Chief Statistician on statistical and methodological issues that affect the quality of Federal data products. The FCSM has several sub-groups, two of which would be of interest to this audience. One is the Geospatial Interest Group (GIG), and another is the Workgroup on Integrated Data Quality. The Workgroup on Integrated Data Quality held a series of workshops, including one focusing on data quality issues associated with the integration of geospatial data. I will give a brief summary of the key outcomes and issues from these workshops, as they relate to the theme of this summit. I will then provide examples from the Bureau of Labor Statistics and the Bureau of Transportation Statistics, where data products were created by integrating data from different sources. I will conclude the talk by briefly discussing an important issue related to geospatial outcomes and products: data ethics.

Presentation

The AI Tsunami

Philippe Cases, ReadWrite Labs

Abstract

AI is a game changer and appear to be one more than any other technology ever. The adoption at which the industry is adopting the technology is mind boggling to the point that we don't talk about technology wave anymore but about an AI tsunami. As an example. 40% of the top performer CIOs believes that AI is a game changer for their organization in 2019. This is 3 times more CIOs than ones who believes that Cloud is a game changer.

If the fundamentals are strong, AI will be adopted based on ROI and few verticals or use cases are growing faster than others. In addition, the amount of data necessary to create and maintain AI systems are very important and costs associated with processing and storing the data are not insignificant, which creates a requirement for more optimization and even an architecture, away from cloud only, edge computing.

We will review where AI is today, the different use cases that drives the adoption of AI and we will review the architecture necessary to mine AI to its fullest. We will also look at the datasets necessary for AI to be successful including location data.

Presentation

Fundamental Issues in Geospatial Data Science: Emerging Trends in Data and Analytics

Marc P. Armstrong, The University of Iowa, marc-armstrong@uiowa.edu

Abstract

Advances in geospatial data capture technologies are fundamentally changing the way that GIS applications are conceptualized. In the past, most applications were developed assuming static information or “data at rest”. This is increasingly not the best option. Massive quantities of geospatial data are now being streamed in real time by, for example, IoT sensors, telematics, UAVs and CubeSats. Taken alone, the volume of this data poses computational challenges for many types of analyses. While that is bad enough, data steam velocities overwhelm those methods that have been developed over decades to populate conventional GIS toolboxes. Even worse, these data problems are occurring when hardware performance has begun to plateau with the suspension of Moore’s Law. In this presentation I discuss emerging streaming data sources and briefly review approaches to parallel analysis of geospatial data, particularly those employing distributed cyberinfrastructure. Though parallelism can address some volume issues, entirely new classes of sampling strategies and analytical methods must be developed to support the geospatial analysis of high-velocity data streams. These new types of analyses must be developed with an eye on emerging hardware trends including neuromorphic and heterogeneous architectures. And then, of course, there is the polarizing quantum quandary. But there is good news: there are plenty of interesting challenges ahead.

Presentation

Building a Generation of Government Data Scientists

Jeanne Holm, City of Los Angeles

Abstract

The City of Los Angeles leads the Data Science Federation that brings together 88 cities and students and professors from 18 universities to work on "homework for their hometown." These students are paired with City departments that have real data and real challenges ranging from ending homelessness to improving air quality to reducing traffic and use open data and GIS to create innovative, sustainable solutions.

Presentation

Knowledge-Powered Data Science for Integrated Modeling in Geosciences

Yolanda Gil, USC Spatial Sciences Institute

Abstract

Major geoscience challenges are presented by complex phenomena with uncertain, intermittent, sparse, multi-resolution, and multi-scale data. New approaches are needed that leverage the advances in data-driven research with methods that exploit the domain knowledge and scientific principles that govern the phenomena under study. These geoscience-aware systems will need to incorporate extensive knowledge about phenomena that combine physical, geological, chemical, biological, ecological, and anthropomorphic factors. I In this talk, I will discuss a research agenda for significantly more powerful data science approaches to sensing and data collection, data integration, modeling, and decision making. I will present our ongoing work on interdisciplinary model integration spanning climate, hydrology, agriculture, and socioeconomic modeling, where we use extensive knowledge about data and models to significantly reduce the time needed to develop new integrated models that currently require months of effort.

Presentation

Scaling machine learning to handle visual data will result in more powerful AI

Nils Lahr, CEO Orions Systems

Abstract

Video Analytics, Machine Learning, and Big Data Analytics are all coming together to create a new form of Artificial Intelligence. In order to address the volume, veracity, and live-streams of visual data, the industry is having to abandon its mature structured data solutions. Additionally, unlike abstract financial and text-based data, “visual” data contains information about real-world activities. Being able to extract activities and actions from multiple sensor types, such as cameras and IoT devices, will allow us to build accurate digital-twins or a virtual representation of the real world. To build these models, we need more innovation around how to map the output of visual machine learning models back into the temporal and spatial domains. Once we have such accurate ‘twins’, we can start to predict the future, understand the world at a macro level, and empower humans with superhuman abilities.

Presentation

Perspective: Harnessing the Power of Data to Support Community Health and Well-Being

Stephanie Shipp, U. of Virginia

Abstract

Local communities would like to use evidence for policy to improve the social determinants of health of our communities through data and data science methods. The University of Virginia’s Social and Decision Analytics Division’s partners with communities to address their pressing local issues. Our innovations apply statistical and social science methods to create CommunityScapes. These quantitative characterizations of a community’s health and well-being are at geospatial levels that matter to decision makers (e.g., neighborhoods, school attendance areas, political districts). This community-driven approach leverages the power of data science to better understand the context in which communities live, learn, work and play to inform policy.

Presentation

When HPC met AI - Next generation of Geospatial Intelligence powered by the ABCI

Satoshi Sekiguchi
Vice-President, National Institute of Industrial Science and Technology, Japan
s.sekiguchi@aist.go.jp

Abstract

Artificial intelligence (AI) is a disruptive technology that can bring benefits to our daily lives, cooperating with IoT, Big Data, Robots, advanced biotechnologies, etc. It has become a powerful force driving innovation across all sectors, including healthcare, manufacturing, transportation, agriculture, and security. However, recent AI algorithms in today's 3rd boom of AI, such as deep neural networks, require a significant volume of training data and high-performance computing resources for practical applications. In this talk, the author highlights contributions from HPC technology that is indispensable for learning at a practical level and the importance of the cooperation between HPC and AI in future geospatial intelligence. More specifically, the author introduces ABCI (AI Bridging Cloud Infrastructure) as a new infrastructure of the world's largest AI-focused, open, and public computing infrastructure for practical AI. ABCI has been operational since 2018 and designed as a reference architecture for data centers combining HPC and AI to discover innovations in the future. The ABCI has been supporting not only industrial AI applications deployed for use in society, but also academic AI research. One of the first grand challenge projects we are supporting is mapping whole brain-scale signal integration events in zebrafish. The author also addresses challenges to provide real-time geospatial intelligence based on various sensor datasets, and actions we have to take for the future.

Presentation

The Age of Precision Conservation: Applying AI and Collaborative Science to Prevent Species Extinctions

Regan Smyth
NatureServe

Abstract

NatureServe is a biodiversity information network that spans the western hemisphere, producing essential information about species and ecosystems, where they are found, and the degree to which they are imperiled. Advances in data science are transforming the way we work - and the quality of data available to guide conservation. One example of this is our habitat suitability modeling initiative, which combines 40 years of field inventory data, advanced spatial modeling techniques, high-powered computing, and the biological expertise of our network to map habitat for over 2,000 of our nation’s most imperiled species and identify priorities for conservation. The success of this and similar efforts hinges on continued investment in quality-controlled data inputs and partnership between the technology and non-profit sectors.

Presentation

Analytics And Representations in SQL and GQL

Keith Hare

High-level outline:

Presentation

Earth Observation data and analytics supporting policy and geospatial industries

Patrick Griffiths

Abstract

Free and open Earth Observation (EO) data available today is unprecedented in terms of quality and quantity. The European Space Agency together with the European Commission operates the Sentinel fleet of EO satellites as part of the Copernicus program. The currently operational 7 satellites from four missions acquire 25 TByte/day of new data while ESA disseminates 250 TByte/day. At the same time “new space” actors are building large EO cube/small sat constellations with often impressive imaging capacities.

This wealth of EO data has triggered fundamental changes in the way we store, process and analyse EO data. Users are increasingly demanding “analysis ready data” that is hosted in and co-located with powerful distributed compute environments and that can be easily accessed at the per-pixel level. Data cubes and cloud-based platforms for processing and analyses allow for the direct employment of methods and technologies from the broader data science domain. Virtually all application areas of EO have benefitted from higher quality and quantity of data. Especially the unprecedented temporal repeat frequencies allow for greatly improved information retrievals.

One example of a sector that strongly affected by these developments is agriculture. The Common Agricultural Policy (CAP) of the European Union (EU) is the single largest budget item in the EU (43% of total budget). EU regulations are now in place that require continuous and complete EO time series based monitoring at parcel level with automated algorithms to ensure compliance of farmers for received subsidies. Currently, systematic environmental impact assessment using Sentinel data is being implemented in the EU legislation. These development are creating a large demand for geospatial services and expertise provided by private companies, are diversifying business models and are increased employment opportunities. Here increased EO data quality and quantity is enabling the implementation and monitoring of major EU policies using algorithms and analytics from the broader data science domain.

Presentation

Spatial data science moves from the avant garde to the mainstream

Lauren Bennett

Abstract

As spatial data science moves from the avant garde to the mainstream, much can be done to help ensure that this new wave of analytics lives up to its promise. It isn’t enough to develop cutting edge algorithms. We also need to build these methods into the GIS ecosystem in such a way as to ensure that they can be applied to a wide range of problems, explained to a broad set of decision makers, and operationalized within organizations so that the knowledge created is used to drive action. GIS has a key role to play, as it can integrate the work being done across disciplines in traditional and spatial statistics, machine learning and deep learning approaches and can put that work into the hands of everyone from scientists to citizens as approachable tools and workflows. To help solve the world’s most complex spatial problems, and make sure the promise of spatial data science is realized, some of the most challenging work will be making sure that the best tools are made simple, approachable, and easily integrated and extended.

Presentation

Data science, an interdisciplinary approach

Anand Kannan

Abstract

Devices generate more data than ever. Organizations collect more data than ever, not always with a clear plan on usage. There is an emerging body of knowledge that enables analysis of such data to identify patterns (unsupervised machine learning), and associate outcomes with input variables (supervised learning). Prediction is getting inexpensive with the combination of such data, analysis techniques and a decline in costs of storage and computational power.

The combination of data science and location intelligence provides us exciting possibilities on how we can use data to solve real problems. This requires thinking in an inter-disciplinary fashion. In many organizations, successful teams have members with expertise in location intelligence, statistics, data science, deep learning and big data. Such teams expand the range of solvable problems for the organization. On the business side, organizations need to develop skills to recognize their internal data strengths, learn of the technical possibilities, and pick problems with significant business impact.

Such a requirement for cross disciplinary thinking is a both an opportunity and a challenge for geospatial organizations. There is quite a bit of learning and unlearning involved. This is an exciting time both for geospatial enthusiasts and data scientists.

Presentation

Designing the Future of Data Science

Anand Kannan

Abstract

Data science is the art and craft of people leveraging technology to create value out of data. We have ever more data, computing power, and analytical techniques. The opportunities are immense, as are the stakes. What’s the future of data science, and why does that future matter? This talk draws on Andy’s experiences advising, creating, and leading industry and government organizations to show how we can better design that future.

Presentation

New opportunities through big mobility data analytics

Kathleen Stewart
Center for Geospatial Information Science
University of Maryland, College Park, MD

Abstract

We all travel a great deal, whether it is in the context of our families and social ties, or for our jobs. Our daily choices regarding where we go and how we get there, contribute to a much larger picture of daily travel patterns. Much of our movement is routine and regular but there are also occasions when travel patterns vary, for example, for a major sporting event or during a flood or hurricane hazards when evacuations occur. With the increasing availability of smartphones and location-based services apps, massive mobile device location data is being generated containing detailed information about trips and travel routes. These data offer great potential for providing insights and a deeper understanding of mobility. My panel talk will consider these emerging and rapidly expanding data sources and what challenges lie in harnessing these data and extracting meaningful information about mobility patterns.

Presentation

Geospatial Intelligence through Accelerated Analytics & Data Science at Scale

Todd Mostak

Abstract

The explosion of IoT data from sensors, smart phones, social media, and other special collection assets provides vital spatio-temporal context, especially when combined with additional big, non-geospatial data sets. New opportunities are arising for government agencies to leverage CPU or GPU hardware to analyze and visualize these massive data sets at interactive speeds. On this panel, we'll discuss how accelerated analytics can be harnessed to rapidly extract geospatial intelligence from big data, and how this technology is enabling businesses and government to take faster action and make more informed decisions.

Presentation

AI-IOT and Location

Milind Naphade

Abstract

The explosion in IOT is creating a flood of big and fast streaming data with spatial information and what are the new frontiers of spatial computing to solve real world AI-IOT problems. NVIDIA is creating – cuSpatial and Metropolis - in the context of applications such as city-scale traffic monitoring, smart garages, etc.

Presentation

CyberGIS-Jupyter for Reproducible Geospatial Research and Education

Anand Padmanabhan

Abstract

Geospatial research and education have become increasingly computation-, data-, and collaboration-intensive. There is also a growing recognition for solutions to bridge the significant digital divide between advanced cyberinfrastructure and geospatial communities and make it accessible to broad geospatial community. In this context, cyberGIS has also emerged as new-generation geographic information science and systems (GIS), based on seamless integration of high-performance and distributed computing, modeling, spatial analysis, visualization, and collaborative problem-solving capabilities. My panel talk will introduce recent advances of cyberGIS with a particular focus on CyberGIS-Jupyter that uses Jupyter notebooks for conducting reproducible geospatial research and education based on cutting-edge cyberinfrastructure and cyberGIS capabilities.

Speaker

Lauren Bennett

Bio

Lauren Bennett leads Esri’s spatial statistics software development team. She provides vision for analytical tools in the areas of spatial statistics, machine learning, and spatiotemporal analysis, and is involved in the R&D and validation of educational resources for new tools. Lauren received a B.A. in geography from McGill University, an M.S. in geographic and cartographic science from George Mason University, and a Ph.D. in information systems and technology from Claremont Graduate University.

Speaker

Yolanda Gil

Bio

Dr. Yolanda Gil is Director of Knowledge Technologies at the Information Sciences Institute of the University of Southern California, and Research Professor in Computer Science and in Spatial Sciences. She is also Director of the USC Center for Knowledge-Powered Interdisciplinary Data Science. She received her M.S. and Ph. D. degrees in Computer Science from Carnegie Mellon University, with a focus on artificial intelligence. Her research is on intelligent interfaces for knowledge capture and discovery, which she investigates in a variety of projects concerning scientific discovery, knowledge-based planning and problem solving, information analysis and assessment of trust, semantic annotation and metadata, and community-wide development of knowledge bases. Dr. Gil collaborates with scientists in many domains on semantic workflows and metadata capture, social knowledge collection, computer-mediated collaboration, and automated discovery. She is a Fellow of the Association for Computing Machinery (ACM), and Past Chair of its Special Interest Group in Artificial Intelligence. She is also Fellow of the Association for the Advancement of Artificial Intelligence (AAAI), and was elected as its 24th President in 2016.

Speaker

Todd Mostak

Bio

Todd Mostak is the CEO and co-founder of OmniSci. Todd originally conceived of the idea of a GPU-accelerated analytics platform while conducting graduate research at Harvard on the role of social media in the Arab Spring, after tiring of waiting hours or sometimes days for traditional CPU-based platforms to run analytic workflows over hundreds of millions of tweets. He later joined MIT’s CSAIL as a research fellow, under the supervision of Sam Madden and Turing Award winner Michael Stonebraker, before founding OmniSci.

Speaker

Ed Strocko

Bio

Ed Strocko is the Director of the Office of Spatial Analysis and Visualization at the USDOT Bureau of Transportation Statistics, or BTS, a principle federal statistical agency. Ed leads the development of geospatial information and visualization tools including the National Transportation Atlas Database, and directs spatial and network analyses that facilitate the understanding of major issues and inform investment and policy decisions. He employs high quality cartography and innovative web applications to produce relevant, high quality, timely, comparable, complete, and accessible geospatial products and statistical visualizations.

Speaker

Ajay Gupta

Bio

Ajay Gupta is the CEO of Health Solutions Research (HSR) and has been working in the health IT field for the past 4 years focusing on leveraging geospatial tech to merge, analyze, and discover the relationship between health outcomes & costs and social determinants of health. He supports health systems efforts to address health challenges, increase access to quality, affordable healthcare by responding to social factors that inhibit such access. In addition to his work at HSR, Ajay Chairs the Health Domain Working Group for the Open Geospatial Consortium, the global standards setting body for all things Geo. Ajay is also the Secretary of the Board of Directors of Holy Cross Health, a multi-hospital social safety net health system in Montgomery County, MD.

If he isn't using HSR's geospatial systems to map disease rates against social factors to identify solutions to population health challenges, such as the opioid epidemic or maternal mortality, you can find him at sporting events or behind the mic busting ear drums while belting out his favorite pop music tunes from U2, Sting, George Michael, or Imagine Dragons.

Speaker

Anand Kannan

Bio

Anand Kannan has a Master’s degree in Computer Science. He is a Fellow in the Tech Ladder, an elite forum of technicians in Pitney Bowes. He has been working in the Location Intelligence industry for almost three decades. His areas of interests include Deep Learning and spatial databases.

He manages the development of server products in location intelligence, and has been driving the vision behind the use of spatial analysis techniques in conjunction with Machine Learning. His recent work includes the use of Deep Learning for Location Intelligence problems such as site selection and map display.

Speaker

Patrick Griffiths

Bio

Patrick has a background in Geography and Geosciences and a PhD in Earth Observation and Land System Science. He has been a co-PI on the NASA/USGS Landsat Science Team since 2012. After leaving academia, he worked for the European Commission's Joint Research Centre (Italy) in the Food Security Unit, involved in agricultural monitoring of food insecure countries and in the context of the EU's Common Agricultural Policy. In 2018 he took a position with the European Space Agency (ESA) in the Earth Observation Directorate in Fracati, Italy. His work there involves the development and support of digital platform technologies for Earth Observation data and analytics, the implementation of innovative EO applications with European Industry and activities dedicated to new machine learning and in-situ/reference data handling.