Sanghani Center leads collaborative study to improve both discovery and traceability of illegally-sourced timber

Reference sample collections from World Forest ID

Virginia Tech has received funding from the National Science Foundation for a collaborative research project that brings machine learning and data science research to the domain of Stable Isotope Ratio Analysis (SIRA) to improve discovery and traceability of illicitly-sourced timber products. Illegal timber trade (ITT) is the most profitable natural-resource crime, valued at 50-152 billion U.S. dollars per year.

Naren Ramakrishnan, the Thomas L. Phillips Professor of Engineering and director of the Sanghani Center for Artificial Intelligence and Data Analytics, is serving as principal investigator for the project with the University of Washington, World Forest ID, and Simeone Consulting, LLC.

“To enforce timber regulations and international frameworks, there is a need for accurate, cost-effective, and high-throughput tools that can be used to identify and trace illegally sourced timber products,” Ramakrishnan said. 

The team brings together data scientists, analytical chemists, geospatial and remote sensing scientists, practitioners, international trade and supply chain specialists, and field experts who conduct reference sample expeditions to bring novel data science approaches to analyzing a range of geospatial and remotely sensed datasets.

Patrick Butler, senior research associate, and Brian Mayer, research associate at the Sanghani Center will be part of the Virginia Tech team.

Key foci of this project include machine learning methods for SIRA analytics; location determination from isotopic ratios; and active sampling strategies to close the loop. Foundational machine learning contributions in science-guided machine learning, contrastive and generative learning paradigms, and active sampling algorithms will support not only the specific domain of SIRA but other adjacent domains in environmental conservation, agricultural forecasting, and smart farm modeling. 

“For example, what we learn from our research could be directly applicable to tracing many other illicitly-sourced products and product inputs, including forest risk commodities such as cocoa, soy, and beef,” said L. Monika Moskal, professor at the University of Washington.

The study will have broad and far-reaching impacts on American security and prosperity, as well. 

“Many key U.S. adversaries rely on illegal logging to finance their activities,” said Jade Saunders, executive director at World Forest ID. “Detecting and curbing such activities will moderate sources of regional instability and threats to U.S. interests.”

The project will lead to improving geospatial prediction accuracy of product origin and will enable a cost-benefit analysis to minimize future data collection costs and optimize prediction gain. Finally, this project will also positively affect U.S. economic competitiveness by reducing competition with illicit actors and moderating risks to international trade, Ramakrishnan said.

Human-Centered Future of Work Symposium set for Nov. 3

Sue Ge, director of ICAT’s Center for Future Work Places and Practices, addresses faculty at the center’s spring networking event. Virginia Tech photo

As technology continues to revolutionize industries and alter the nature of everyday life, the future of work can seem unclear. Virginia Tech’s Institute for Creativity, Arts, and Technology (ICAT) is bringing together a wide breadth of expertise to discuss this topic during the Human-Centered Future of Work Symposium.

Sponsored by the Department of Economics and the Kohl Center, AAEC, the symposium feature a policy roundtable discussion that aims to search for the common ground on the human-centered future of work.

One of the panelists is Chris North is a professor of computer science at Virginia Tech and the associate director of the Sanghani Center for Artificial Intelligence and Data Analytics. Read full story here.

New Data and Decision Sciences Building encourages collaboration to address world’s data challenges

The Data and Decision Sciences Building. Photo by Noah Alderman for Virginia Tech.

Virginia Tech’s new Data and Decision Sciences Building has opened its doors to students, faculty, staff, and industry professionals ready to tackle some of the world’s most pressing data challenges. Completed in the summer, the 120,000-gross-square-foot facility houses multiple colleges including the Pamplin College of BusinessCollege of Engineering, and College of Science.

Several faculty from the Department of Computer Science have offices located in the building, along with labs and classrooms that allow students to experience and interact with the latest computational technologies. The new visualization lab features a high-resolution power wall with multi-touch functionality. Coupled with SAGE3 software developed by researchers in the Sanghani Center for Artificial Intelligence and Data Analytics under a $5 million dollar National Science Foundation grant, the high-resolution screen enables the display and organization of large amounts of media, data analytics, and visualizations. Read full story here.

Sanghani Center Student Spotlight: Syuan-Ying Wu

Poster for published paper “MetroScope: An Advanced System for Real-Time Detection and Analysis of Metro-Related Threats and Events via Twitter”

Metro systems are vital to many people’s daily lives, but they face safety or reliability challenges, such as criminal activities or infrastructure disruptions. Real-time threat detection and analysis are crucial to ensure their safety and reliability. 

Syuan-Ying (Justin) Wu, a master’s degree student in computer science whose research focuses on social media analytics and software development, is currently part of a research team that is working with the Washington Metropolitan Area Transit Authority (WMATA) to address these issues.  

With fellow students at the Sanghani Center and his advisor, Chang-Tien Lu, Wu has been instrumental in developing the MetroScope real-time threat/event detection system that can automatically analyze event development; prioritize events based on urgency; send emergency notifications via emails; provide efficient content retrieval; and self-maintain the system.

“This is a great improvement over many existing systems that can detect the event but cannot analyze it or prioritize it,” Wu said. “And our system offers other advantages like not having to continuously monitor system notifications.”

Their collaborative paper, “MetroScope: An Advanced System for Real-Time Detection and Analysis of Metro-Related Threats and Events via Twitter,” was published in the proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval held in Taipei, Taiwan, this past summer.

Wu, who earned a bachelor’s degree in applied mathematics at Fu Jen Catholic University in Taiwan, said this research collaboration with a metropolitan metro system is a good example of what led him to pursue his master’s degree at Virginia Tech and the Sanghani Center. “The exceptional computer science program and distinguished professors have offered me the opportunity to find ways of applying cutting-edge technology to tackle a real-world problem,” he said. “It has been the perfect environment to achieve my goals.”

Projected to graduate this fall, Wu hopes to secure a position as a software engineer. 

Ming Jin receives NSF grant to introduce antifragility into power systems

Ming Jin

Ming Jin, an assistant professor in electrical and computer engineering and core faculty at the Sanghani Center has received a National Science Foundation grant to revolutionize the design of learning-enabled, safety-critical systems, with a special focus on power systems.

The grant was awarded under the Safe Learning-Enabled Systems (SLES), a partnership between the NSF, Open Philanthropy, and Good Ventures.

Jin will collaborate with Javad Lavaei, professor in Industrial Engineering and Operations Research at the University of California Berkeley.

The project introduces antifragility, a concept that goes beyond robustness which can be compared to a sturdy structure that remains unyielding in a storm but does not grow or adapt from the experience; or resilience which is like a rubber band: when stretched, it can recover by going back into its original shape. 

“We are not merely designing systems to withstand challenges of rare and unpredictable events, but to flourish because of them,” Jin said. 

The task of preserving end-to-end safety of the power system will be crucial, Jin said, though it is complex amidst distributional shifts, driven by the growing complexity and unpredictability of the environment. 

The project will addresses safety challenges through three interconnected research thrusts. The first thrust targets the creation of proactive, antifragile systems that anticipate and adapt to changes, using advanced techniques such as meta-safe learning and offline reinforcement learning. The second thrust bolsters system antifragility through multi-agent systems, encouraging exploration, cooperation, and distributed control to ensure resilience and safety, even under significant disturbances. The third thrust is devoted to validation and stress testing, employing multi-objective adversarial learning and real-world case studies to better handle rare or unexpected events.

“Our algorithms are more than just learners; they’re evolvers. By turning continual threats into avenues for enhancement, we are redefining what safety in power systems looks like,” he said.

Four students advised by Jin will work with him on the project: Vanshaj KhattarAhmad Al-TawahaZain ul Abdeen, andBilgehan Sel.

Amazon-Virginia Tech Initiative announces support for two Amazon Fellows and five faculty-led projects for 2023-24 academic year

The Amazon Fellows are (from left) Minsu Kim and Ying Shen. Photos courtesy of the subjects.

The Amazon–Virginia Tech Initiative for Efficient and Robust Machine Learning will support two Amazon Fellows and five innovative research projects led by Virginia Tech faculty in the 2023-24 academic year that further the initiative’s mission of advancing innovation in machine learning. 

The initiative, launched in 2022, is funded by Amazon, housed in the College of Engineering, and directed by researchers at the Sanghani Center for Artificial Intelligence and Data Analytics on Virginia Tech’s Blacksburg campus and at the Virginia Tech Innovation Campus in Alexandria. 

An open call for fellowship nominations and faculty projects went out across the Virginia Tech campuses. An advisory committee of Virginia Tech faculty and Amazon researchers selected two Amazon Fellows from 27 nominations — more than double what was received last year — and five faculty projects from 17 submitted proposals. Read full story here.

Faculty ‘cautiously optimistic’ about the potential of generative AI

Faculty members are learning that generative AI tools are capable of many things: writing essays and emails, customizing lessons and learning, even creating seemingly original art, like this impressionistic painting of a laptop. Illustration created by Melody Warnick using AI.

Faculty are considering how AI models such as ChatGPT can customize learning by producing dynamic case studies or offering instant feedback or follow-up questions. Many are making AI the subject of assignments. They’re asking students to analyze and identify weaknesses in arguments produced by ChatGPT, for instance, or to edit an AI-produced essay with “track changes” on.

That kind of critical thinking about generative AI is vital, said Ismini Lourentzou, assistant professor of computer science in the College of Engineering and core faculty at the Sanghani Center for Artificial Intelligence and Data Analytics. “It’s our responsibility as educators to teach students how to use these tools responsibly, and then understand the limitations of these tools.” Read the full story here.

Sanghani Center Student Spotlight: Vanshaj Khattar

Graphic is from the paper ” Winning the CityLearn Challenge: Adaptive Optimization with Evolutionary Search under Trajectory-based Guidance”

Vanshaj Khattar, a Ph.D. student in electrical engineering, is passionate about use-inspired research and solving real-world problems. 

“More specifically, I am interested in how we can design trustworthy reinforcement learning algorithms that are safe, robust, explainable, and can continually adapt to non-stationarity in the real world,” said Khattar, who is advised by Ming Jin.

Currently, he is working on an offline reinforcement learning (RL) problem for building energy management, where the learning agent has to learn optimal actions from a dataset without access to the environment. 

“Offline RL is hard because not all possible cases are covered inside the dataset,” Khattar said. “I am addressing this partial coverage by proposing an implicit actor-critic method for offline RL using optimization-based policies with a special robustness property to learning errors in offline RL which I am able to exploit to achieve a good performance on a multiple-building energy management problem. At the same time, I am maintaining some key aspects of interpretability which are lacking in current approaches.” 

Khattar earned a bachelor’s of technology degree in electrical and electronics engineering from Delhi Technological University, India, and earned a master’s degree in electrical engineering from Virginia Tech.

While in the master’s program, he worked on motion prediction/planning for autonomous vehicles and came across reinforcement learning methods and their huge successes in many domains such as AlphaGo. 

“However, I realized that the potential of RL methods was mostly being utilized in simulated domains, and real-world applications were still limited,” Khattar said. “This inspired me to pursue my research in building trustworthy RL algorithms that can be applied to real-world applications with safety guarantees.”

He said the opportunity to collaborate was a major factor in attracting him to Virginia Tech and the Sanghani Center.

In 2023, he has presented three papers: “Winning the CityLearn Challenge: Adaptive Optimization with Evolutionary Search under Trajectory-based Guidance” and “On Solution Functions of Optimization: Universal Approximation and Covering Number Bounds,” both at the 37th AAAI Conference on Artificial Intelligence; and “A CMDP-within-online framework for Meta-Safe Reinforcement Learning,” at the International Conference on Learning Representations(ICLR).

Khattar is projected to graduate in 2026 and hopes to continue his research as an industry professional. 

Sanghani Center Student Spotlight: Wenjia Song

Graphic is from the paper “Subpopulation-specific Machine Learning Prognosis for Underrepresented Patients with Double Prioritized Bias Correction”

Cyberattacks have led to substantial losses for both businesses and individual users in recent years raising an urgent need to strengthen protection against such threats,” said Wenjia Song, a Ph.D. student in computer science who is working to address the problem.

“My research focuses on machine learning application and methodology development for improving accuracy on crucial detection problems, including medical predictions and threat detection in cybersecurity, through quantitative experiments,” she said.

Song’s current project is aimed at cyber threat detection. 

“Real-world examples of such attacks include the Colonial Pipeline ransomware attack and the SolarWinds hack,” she said. “We try to detect malicious attack behaviors at an early stage in order to minimize the damage they may cause.”

Prior to entering the doctoral program, Song earned two bachelor of science degrees – in computer science and in mathematics — from Virginia Tech. 

“As a Ph.D. student, I was attracted to the Sanghani Center because of its reputation for diverse and impactful research,” she said. “I really like being part of a thriving academic community where I receive significant encouragement and support from both professors and my peers.”

Song is advised by Danfeng (Daphne) Yao.

Among her published papers are: “Subpopulation-specific Machine Learning Prognosis for Underrepresented Patients with Double Prioritized Bias Correction,” in Communications Medicine in 2022; and “Specializing Neural Networks for Cryptographic Code Completion Applications,” in IEEE Transactions on Software Engineering in 2023.

Song also presented her work on measurement of ransomware behaviors and evaluation of defenses at both the Commonwealth Cyber Initiative (CCI)/Virginia Tech Transportation Institute (VTTI) Tech Showcase in 2022 and at the CCI Symposium in 2023.

In 2022, she presented the poster “APT Detection through Sensitive File Access Monitoring” at the Network and Distributed System Security (NDSS) Symposium and the poster “Behavioral Characterization of Crypto-Ransomware and Evaluation of Defenses” at the IEEE Secure Development Conference. 

She also gave a lightning talk, “Crypto-ransomware Detection through Quantitative API-based Behavioral Profiling,” at USENIX Security 2023. 

Projected to graduate in May 2024, Song said she would like to continue her research in an industry position.

Making a CAREER on bridging scientific knowledge and AI

Anuj Karpatne. Photo by Peter Means for Virginia Tech.

Anuj Karpatne,
associate professor in the Department of Computer Science in the College of Engineering has won a five-year, $595,738 National Science Foundation Faculty Early Career Development Program CAREER award to explore a unified approach for accelerating scientific discovery using scientific knowledge and data. Karpatne is also a core faculty member at the Sanghani Center for AI and Data Analytics. Read the full story here.