LMI Research Institute Academic Programs Newsletter | Volume 2

Our five academic partners help LMI integrate academic rigor into our independent research and development, offerings, and capability development to support our government clients. Through our vital collaboration, we stay ahead of customer challenges and help customers anticipate and clear upcoming hurdles.

LRI Academic Programs Names LMI Liaisons for Partner Universities

Along with their responsibilities in their respective service lines and markets, these liaisons build and strengthen LMI’s presence and influence with our academic partners.

In addition to our academic partners, LMI leverages more than 30 additional universities, including Carnegie Mellon University, George Mason University, Georgia Tech, Harvard University, the Massachusetts Institute of Technology, Morgan State University (HBCU), North Carolina Agricultural and Technical State University (HBCU), Purdue University, Tennessee State University (HBCU), UC Berkeley, the University of Missouri, Virginia State University (HBCU), Virginia Tech, and William & Mary, on over 60 contracts and opportunities.

Contact an academic liaison for more information on how you can be involved with our academic partners. Contact Linda Bixby, PhD, director, LRI Academic Programs, for more information on our LRI Academic Programs efforts and initiatives.

Linda Bixby, Director, Academic Programs

lbixby@lmi.org, 571-209-2587

Innovating with NASA

The Aftershock: NASA Shock Propagation Prediction Challenge looks for novel shock propagation prediction models that improve NASA’s ability to predict shock loads through spacecraft. NASA currently utilizes techniques from the 1970s and hopes to improve the accuracy and versatility/extensibility of the propagation models while maintaining usability (time and cost).

Launched on January 26, 2022, the Aftershock: NASA Shock Propagation Prediction Challenge is a single-phase competition in which participants submit their solutions outlining their methodology and data findings.

Who should enter? If you have experience in the following disciplines, then you're a perfect match:

  • Mechanical Engineering
  • Physics & Astroscience
  • Data science
  • Or related disciplines

Winners are eligible for a share of US$50,000 of cash prizes and will receive a certificate of achievement from NASA.

If you’re interested, check out the link below to get all the details on the Aftershock: NASA Shock Propagation Prediction Challenge and sign up today!

A recording of the webinar that accompanied the launch of this challenge can be found here for additional details on the challenge. 

Howard University

Howard University Expands Research Week to Research Month

  • In Summer/Fall 2022, LMI will fund Howard University graduate and undergraduate students and professors to support two LRI research efforts: the Diversity, Equity, Inclusion, and Accessibility Opportunity Metric and Assessment tool and the Genetic Sequence Data Analyses project.
  • Dr. Linda Bixby and Dr. Kim Barnette met with Dr. Danda Rawat, Director of Howard’s Data Science and Cybersecurity Center, and the Department of Defense Center of Excellence in Artificial Intelligence and Machine Learning at Howard University to discuss future collaboration efforts with LMI.
  • LMI included Howard University in two Federal government client proposal efforts (awards pending).

Penn State

LMI’s Dr. Linda Bixby Named CSCR Advisory Board Vice Chairperson

The Board represents CSCR corporate sponsors and business leaders that use relationships with academia to stay current in this rapidly changing supply chain management landscape. Additional industry representation on this board comes from The Hershey Company, IBM Corporation, Walmart, SDI, Merck, NTT Data Services, and CLX Logistics.

Penn State’s CSCR is one of the nation’s leading institutions dedicated to research and education in supply chain management. With over 30 affiliated faculty members from the Department of Supply Chain and Information Systems at the Smeal College of Business, LMI has access to one of the largest and most prestigious academic concentrations of supply chain management research in the United States.

The Center for Supply Chain Research, part of the internationally ranked Penn State Smeal College of Business, connects students, academics, and professionals from leading organizations within a community that is shaping the future of the supply chain discipline.

Supply Chain Leaders’ Forum

Industry experts discussed the various impacts the pandemic has had on global supply chains and how organizations are responding to the challenges of the future. There were 127 industry participants representing 31 companies as well as Penn State faculty, staff, and students in attendance.

LMI’s Shaye Brotherton, senior principal, was a featured speaker, presenting on Hybrid Work and the Evolving New Normal. Shaye discussed changes in the workplace, how the workforce is evolving and implications for the future, and what those changes mean for supply chain leaders that need to adapt. In a later session, Matt Peterson, LMI supply chain principal and liaison to Penn State, summarized LMI’s approach to sponsoring Penn State student research and then introduced Penn State honors student, Emily Irvin. Emily discussed her LMI-sponsored research project When to Onshore? A Framework for the Manufacturing of Active Pharmaceutical Ingredients.

This student research effort is one of six active LMI-sponsored projects—all linked to LMI offerings, capabilities, or prototypes. Our other current research projects include commercial supply chain technologies, supply chain risk, digital acquisition systems, and climate change logistics. As a CSCR corporate sponsor, LMI submits up to three research topic proposals per semester. Prior to the Fall and Spring semesters, Academic Programs asks LMI staff for ideas for research topics and offers the opportunity to be a project subject matter expert to help guide the student research.

The supply chain leaders’ forum is held in the spring and fall of each calendar year. The forum fosters peer-to-peer discussions of current issues in logistics and supply chain management to seek commonalities in business practice. The invitation-only program is intended for mid-to senior-level supply chain professionals and is one of the signature events on the CSCR calendar.

University of Maryland

Drone Policy Research Continues

UMD Professor Charles Harry, alongside Graduate Researcher Victoria Adofoli, continue to construct a robust research environment to analyze complex Unmanned Aircraft Systems policies on the state and local level. Thus far, the research has provided surprising and insightful results.

Highly regulated states like Texas, Oregon, Nevada, and North Carolina focus on surveillance, public safety, drone operations, and property damage. Victoria Adofoli commented, “I found it surprising. I previously assumed it would be the more progressive states with the most policies. The underlying motivation behind these state policies requires further investigation from future researchers.”

Moderately regulated states like California, Virginia, and Florida emphasize surveillance, environment, public safety, and drone operations. Less regulated states place an emphasis on public safety and surveillance.

LMI Hires UMD Graduate in Suntiva Acquisition

Through LMI’s recent acquisition of Suntiva, we were happy to welcome Suntiva VP of the Defense and Intelligence Business Unit, Scot Stitely, to our team. Scot comes to LMI with 19 years of federal contracting and leadership experience, with a specific focus on corporate growth strategy development and execution, profit & loss management, strategic IT planning and governance, emerging technology adoption, and portfolio, program, and project management. Scot attended University of Maryland form 1999-2003 where he completed his bachelor’s in Economics. Scot is a proud alumnus of UMD and great addition to the LMI team. Welcome Scot!

University of Virginia

Opportunities and Challenges for Indirect Sensing in Smart Buildings

LMI is funding UVA research for a 2022 academic partnership project, “Opportunities and Challenges for Indirect Sensing in Smart Buildings,” that includes the development of a framework and supporting methods for machine learning algorithms on IoT device sensing data (singular or devices in combination) to protect and best manage “smart” buildings. This research builds on a previous LMI-funded IoT effort with UVA, “Creation of a Test Framework and Test Harness to Evaluate IoT Devices,” to investigate application of a testing framework for analysis of alternatives for sensors and devices, including their operating systems.

This research is important to LMI and our clients who seek to leverage IoT and the vast trove of data provided by sensors. LMI is supporting several Department of Defense (DoD) and Department of Homeland Security (DHS) sensor/device programs with technical or software support including Naval Supply Systems Command (NAVSUP); Joint Health Risk Management (JHRM); DHS Countering Weapons of Mass Destruction Office (CWMD); and Joint Enterprise Omnibus Program, Engineering, and Technical Support (JE-OPETS). These programs use or ingest data from Radio Frequency Identification (RFID) sensors; chemical, biological, radiation, narcotics (CBRN) sensors (e.g., Joint Chemical Agent Detector [JCAD]); the DoD Integrated Sensor Architecture (ISA); Biological Pathogen Triggers, Collectors, Identifiers; and others. Additionally, LMI recently proposed a solution to integrate artificial intelligence and machine learning (AI/ML) into a comprehensive, integrated wearable sensor platform for a DoD agency.

Project Overview—Opportunities and Challenges for Indirect Sensing in Smart Buildings

This project aims to identify how different indoor sensors can be used to infer properties of buildings other than what they are intended and designed to monitor. This “indirect sensing” presents an interesting opportunity for smart and instrumented buildings. Typically, observing a new characteristic of a building requires installing a new sensor network targeting that application. By sensing indirectly, we envision that existing sensors can be reused to monitor new conditions as applications and priorities change. However, these techniques can be exploited maliciously, as sensor data intended to monitor one property in a building might expose unintended insights the building operator may be unaware of. Unauthorized access to the data may leak more about the activity in the building than expected.

UVA researchers are currently analyzing existing literature to identify state-of-the-art ML techniques to gather different information from commercially available indoor sensors (other than their intended purpose). Next, they will explore different strategies to enable indirect sensing to compare the strengths and weaknesses of each. Research will ultimately culminate in an ML-based approach to using sensors for other their intended purpose as well as a general framework to guide future “smart building” assessments (including design, construction, and sensor deployment considerations).

UVA Points of Contact:

  • Dr. Brad Campbell, Assistant Professor, Assistant Professor, Computer Science; Assistant Professor, Electrical and Computer Engineering
  • Dr. Arsalan Heydarian, Assistant Professor in the department of Engineering Systems and Environment as well as the UVA LINK LAB

LMI POC:

  • Michael Brennan, Director, Digital Solutions

Capability Expansion through Capstone Collaboration

With the end of the academic year and capstone program rapidly approaching, LMI’s UVA capstone teams are making exciting progress in the refinement of their projects.

Synthetic Data Validation

LMI’s synthetic data validation project with UVA is designed to create a tool for data scientists to validate synthetic data, particularly with respect to accuracy and privacy protection.

Accuracy validation activities focus on developing methods to evaluate whether synthetic data preserves variable behaviors and relationships as one would expect in a “ground truth” data set.

Privacy activities prioritize whether models are able to obfuscate sensitive data attributes as a measure of data reconstruction.

To achieve these objectives, the team has performed significant research into synthetic modeling approaches and scoring/evaluation methods and became familiar with various methodologies for managing data anonymization, privacy, and reconstruction to support the validation process. The final product will include an aggregate quality/“goodness” score to convey the accuracy and privacy strength of a given synthetic data set.

Testing and Evaluation for Analytic Models—Automating Test Generation

LMI’s Testing and Evaluation (T&ES) for Analytic Model project with UVA is designed to produce a tool that is capable of identifying the right data sets and methods of evaluation to test the outputs of machine learning models.

The team working this capstone divided the project into two primary buckets, data acquisition and model development. Data acquisition activities began with an effort to scrape open-source data science to create a database for model training while model development activities primarily explored Naïve Bayes and Bidirectional Encoder Representations from Transformers (BERT) as methods to help inform the formal model development process.

Automated Ontology Development

LMI’s Automated Ontology Development project with UVA is designed to create an automated tool that can process a body of text and learn context-specific synonyms, hyponyms, and hypernyms.

To date, the team has designed a text-processing pipeline for cleaning, segmenting, and parsing documents and created a framework for reading and comparing Resource Description Framework (RDF) ontologies. Near term activities to complete this project include implementing a Hearst Pattern approach to generating hyponyms, capturing results using an RDF framework, and benchmarking them against the baseline RDF ontology.

2022 Women in Data Science Sponsorship

This year, numerous LMI practitioners including Cynthia (Cindy) Goss and Katie Thompson were honored to attend the UVA School of Data Science conference, hosted in Charlottesville, VA as part of the annual WiDS Worldwide conference organized by Stanford University and hosted at an estimated 200+ locations worldwide. In addition to attending the event, Cindy had the privilege of introducing the “Data and Human Rights” panel which showcased leaders from across industry and academia.

The panel included discussions from responsible data use from surveillance and privacy to automated decision systems while emphasizing the increasingly important connection between data science and civil and human rights. This panel also explored efforts to ameliorate the harms of datafication as well as data use in support of justice and rights. Participants included:

  • Megan Price, Executive Director, Human Rights Data Analysis Group
  • Maria De-Arteaga, Assistant Professor, McCombs School of Business & Machine Learning Laboratory, University of Texas at Austin
  • Karolina Naranjo (M.S. in Data Science 2022)
  • Deborah Hellman, David Lurton Massee, Jr., Professor of Law, UVA School of Law

In conjunction with the WiDS conference, and as representatives of the LMI Women in Data Science community, Cindy and Katie speak about what data science means to them and LMI.

University of Texas at Austin

Effect of Building Occupancy on Energy Consumption UT Austin’s CEDA Research Sponsored by LMI

Professors Ben Leibowicz and Eric Bickel along with Kyle Skyllingstad, a student in the ORIE (Operations Research and Industrial Engineering) MS program, are working with LMI to develop, explore, and validate models that describe energy consumption related to employee occupancy in large office buildings. The first part of this research constructs a simple multiple linear regression to shed light on the impacts of selected parameters on building energy usage.

Building parameters in the research model include occupancy levels, structural aspects, and environmental factors. Daily data for these parameters, as well as different types of energy usage, is obtained for a group of candidate buildings selected to model corporate office buildings. The data spans a three-year period which encapsulates the COVID-19 pandemic. By fitting a linear regression model, the research team measures the relative impact of each predictor feature on energy usage reduction. The model is centered around buildings on the UT Austin campus and can be expanded to predict energy usage for federal and commercial buildings with known parameters.

The research team also explores the results from a simulation based on building energy models with stochastic occupancy, dynamic energy use, and historical observed weather characteristics. This simulation records energy use measurements at ten-minute intervals over one year. The research compares and contrasts benefits of developing building management strategies based on simulation model versus observed UT campus building data. Ultimately, this research will support decisions around planning for future facilities requirements and energy costs based on expected building occupancy.

UT and LMI Partner on Business and Recruiting

In the first quarter of 2022, LMI and UT collaborated on three business opportunities across the federal government to support recruiting of UT students for fellowships, internships, and full-time positions. LMI met with UT career centers across campus and attended the virtual career fair sponsored by the College of Natural Sciences. Students interested in developing their careers as part of our diverse workforce should contact Oswaldo Velasquez, LMI Talent Acquisition.

Download

White Papers and Case Studies

Please fill out the form below in order to download the document.
Name
Yes, I’d like to receive future LMI news, trends, and updates. LMI will never sell or distribute your email to any third parties.
Privacy Policy 

Contact Us

Yes, I’d like to receive future LMI news, trends, and updates. LMI will never sell or distribute your email to any third parties.
Privacy Policy