Spring Break Ends

Wednesday / CS Seminar
April 2nd / 1:00 PM
Hybrid / Mudd 3514

Speaker
Yevgeniy (Eugene) Vorobeychik

Talk Title
Achieving AI Safety in a Contested World

Abstract

"As the increasing capabilities of AI-enabled systems have led to broad deployment across diverse applications ranging from conversational agents to self-driving cars, safety considerations have come to be central to the current research agenda. However, the very meaning of safety has come to be broad and in some cases contested. For example, there may be responses to conversational prompts that some may deem neutral, while others offensive, or autonomous driving behaviors that some may view as efficient while others perceive them as dangerously aggressive. A useful way to conceptualize safety considerations is to divide these into two categories: objective and subjective. The former (for example, running over a pedestrian) is not reasonable contested, while the latter (for example, how aggressively a self-driving car should merge onto a freeway) can admit a range of legitimate perspectives.

In this talk, I will present our recent work tackling both objective and subjective safety considerations. On the former, I will present learning-based approaches for synthesizing provably stable and safe neural network controllers in known dynamical systems, combining gradient-based methods for both synthesis and verification with ideas from curriculum learning. Further, I will briefly discuss our recent work that facilitates safety specifications that combine natural language with formal logic, in which we combine LLMs with conformal prediction to obtain provably correct plans. For the latter, I will discuss an axiomatic framework for preference learning that accounts for disagreement in safety preferences, as well as a novel approach for reinforcement learning with diverse task (e.g., safety) specifications that achieves provable performance guarantees and state-of-the-art performance in zero-shot and few-shot settings."

Biography
Yevgeniy Vorobeychik is a Professor of Computer Science & Engineering at Washington University in Saint Louis. Previously, he was an Assistant Professor of Computer Science at Vanderbilt University. Between 2008 and 2010 he was a post-doctoral research associate at the University of Pennsylvania Computer and Information Science department. He received Ph.D. (2008) and M.S.E. (2004) degrees in Computer Science and Engineering from the University of Michigan, and a B.S. degree in Computer Engineering from Northwestern University. His work focuses on game theoretic modeling of security and privacy, adversarial machine learning, algorithmic and behavioral game theory and incentive design, optimization, agent-based modeling, complex systems, network science, and epidemic control. Dr. Vorobeychik received an NSF CAREER award in 2017, and was invited to give an IJCAI-16 early career spotlight talk. He also received several Best Paper awards, including one of 2017 Best Papers in Health Informatics. He was nominated for the 2008 ACM Doctoral Dissertation Award and received honorable mention for the 2008 IFAAMAS Distinguished Dissertation Award.

Research/Interest Areas
ML, game theory
---
Zoom: https://northwestern.zoom.us/j/98890454611?pwd=MIBSzucB4F5Cm1d8tVCCfTMbDhrFbu.1
Panopto: https://northwestern.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=d10d65da-87eb-4f36-9e46-b2a601452849

Drawing on the Northwestern Engineering whole-brain philosophy and leadership in cognitive science, the Master of Science in Artificial Intelligence program would like to invite you to learn more at our upcoming webinar.

Take this opportunity to join Dr. Kristian Hammond, Professor of Computer Science and director of the MSAI program, as he discusses the complexities of this field, and how this newly offered program at Northwestern Engineering will prepare students for a career in artificial intelligence. At the end of the presentation, we will offer an open Q&A where you will be able to have your specific questions answered. You are also welcome to email your questions to us ahead of the session (msai@northwestern.edu).

Related Info

Register Now

Title: "AI-enabled design of novel self-assembling protein nanomaterials".

Abstract: New AI-based tools for protein design are revolutionizing the field. Dr. King will discuss how his group is applying these methods to design novel self-assembling proteins. Like naturally occurring protein assemblies, these materials provide unparalleled homogeneity, addressability, and functionality. However, unlike naturally occurring assemblies, protein design offers a route to tailoring the structure and function of self-assembling proteins with atomic-level accuracy. Dr. King will describe his group's current efforts in this space, which focus on developing methods to move beyond strict symmetry and design increasingly asymmetric and complex protein assemblies. Applications in nanoparticle vaccine design, including the development of the world's first computationally designed protein medicine, will be discussed.

Speaker: Neil King, PhD; Associate Professor of Biochemistry, University of Washington.

Title : Machine Learning, Music, and Medicine

Abstract: Recent developments in machine learning allow for high-resolution and data-driven multimodal approaches to the development and deployment of adaptive technologies and music-based therapeutic interventions for a wide variety of disabilities and diseases. There is a wealth of behavioral data that correlates participation in musical activities to improved lucidity, memory, emotional regulation, social interaction, and overall well-being for patients with neurodegenerative diseases, though the precise mechanism of action is as of yet not well-understood. This talk explores three different nascent research projects aimed at leveraging various integrative approaches to understanding music cognition and creativity to not only better understand the effects of musical stimuli (whether patient-directed or patient-initiated) on arousal, attention, and emotion, but also to develop adaptive technologies that expand access to music-making opportunities and allow patients to extend their participation in these activities through periods of cognitive decline.

Tina Tallon, Ph.D. is an Assistant Professor of Artificial Intelligence and Music Composition; Area Coordinator of Composition at Ohio State University.

Title: Nonlinear tomographic reconstruction via nonsmooth optimization

Members of the NITMB community are invited to join us for Research-In-Progress meetings, an informal venue for members of the NITMB to discuss ongoing and/or planned research.

Vasilis Charisopoulos is a postdoctoral scholar in the Willett Group at the University of Chicago. He is broadly interested in developing numerical optimization methods for machine learning, signal processing and scientific computing. He holds a PhD in Operations Research & Information Engineering from Cornell University. Vasilis was recognized as a Rising Star in Computational and Data Sciences by the UT Austin Oden Institute in 2023.

Learn more about Vasilis Charisopoulos' research and engage in discussion with the NITMB community. Research-In-Progress talks take place on Wednesdays at 3pm at the NITMB office (875 N Michigan Ave., Suite 4010). Snacks and coffee will follow.

Related Info

Abstract

The increasing complexity of transportation networks demands computational architectures that integrate data, algorithms, and scalability in a unified framework. This talk introduces a tensor-based approach to transportation network modeling and optimization, emphasizing the use of layered computational graphs to handle flows, travel times, and system-wide performance. By leveraging layered tensor operations, network flows are redefined as multi-dimensional entities, enabling efficient propagation of origin-destination (OD) flows, path probabilities, and link travel times. This architecture bridges classical optimization techniques with neural methodologies to address challenges in dynamic traffic assignment, vehicle routing, and multimodal transport systems.

Bio

Xuesong (Simon) Zhou is a Professor of Transportation Systems at the School of Sustainable Engineering and the Built Environment, Arizona State University (ASU), Tempe, Arizona. Dr. Zhou's research focuses on developing methodological advancements in multimodal transportation planning applications, including dynamic traffic assignment, traffic estimation and prediction, large-scale routing, and rail scheduling. Dr. Zhou has served as an Associate Editor of Transportation Research Part C, is currently the Executive Editor-in-Chief of Urban Rail Transit, and an Editorial Board Member of Transportation Research Part B. He has also chaired the INFORMS Rail Application Section (2016 and 2025) and currently serves as a subcommittee chair of the TRB Committee on Transportation Network Modeling (AEP40).

Dr. Zhou is the Director of the ASU Transportation+AI Lab, where he is the principal architect and programmer for several open-source packages, including DTALite, NEXTA, and OSM2GMNS, which have collectively received over 100,000 downloads and many system deployments at various metropolitan planning agencies and state DOTs. He has published over 100 papers in Transportation Research Part B, Transportation Research Part C, and other leading transportation journals, with an H-index of 60 and a total of 11,000 citations in Google Scholar.

In addition to his academic achievements, Dr. Zhou is passionate about connecting practitioners, researchers, academics, students, and others involved in transportation planning and travel modeling. He serves as the conference chair for the TRB Innovations in Travel Analysis and Planning Conference in 2023, and a board member of Zephyr Foundation, a non-profit organization dedicated to advancing transportation research and education.

Related Info

Foundation Models and Generative AI for Medical Imaging Segmentation in Ultra-Low Data Regimes

Pengtao Xie, Assistant Professor, Electrical and Computer Engineering, University of California, San Diego

Abstract: Semantic segmentation of medical images is pivotal in disease diagnosis and treatment planning. While deep learning has excelled in automating this task, a major hurdle is the need for numerous annotated masks, which are resource-intensive to produce due to the required expertise and time. This scenario often leads to ultra-low data regimes where annotated images are scarce, challenging the generalization of deep learning models on test images. To address this, we introduce two complementary approaches. One involves developing foundation models. The other involves generating high-fidelity training data consisting of paired segmentation masks and medical images. In the former, we develop a bi-level optimization based method which can effectively adapt the general-domain Segment Anything Model (SAM) to the medical domain with just a few medical images. In the latter, we propose a multi-level optimization based method which can perform end-to-end generation of high-quality training data from a minimal number of real images. On eight segmentation tasks involving various diseases, organs, and imaging modalities, our methods demonstrate strong generalization performance in both in-domain and out-of-domain settings. Our methods require 8-12 times less training data than baselines to achieve comparable performance.

Related Info

Title: "Opening the Human Blackbox in AI-Assisted Decisions"

By Jessica Hullman, Ginni Rometty Professor of Computer Science and IPR Fellow.

This event is part of the Fay Lomax Cook Spring 2025 Colloquium Series, where our researchers from around the University share their latest policy-relevant research.

Please note all colloquia this quarter will be held in-person only.

Large Language Models (LLMs), like those powering ChatGPT, are increasingly popular and pervasive. Are you interested in using them for your research but unsure how to start? This workshop will introduce research tasks where LLMs can be leveraged effectively. In addition, the workshop will help you understand the differences between the large variety of available models and choose one best suited to your research project.

Prerequisites: None

Register Now

The Slippage: 3D Humanities Series aims to nurture the formation of an interdisciplinary learning community that critically engages the connections between performance, history, theater, and emerging technologies. The 3-part performance and workshop series is designed to expose Northwestern University audiences to professional artists with technology-enhanced practices, and to provide participants with the skills and on-campus resources that enable innovative humanities research. Each event will pair a performance and discussion led by a technology-focused professional performance artist, with a hands-on skills-based workshop led by Northwestern University faculty or staff. The workshops will allow participants to engage technologies such as: motion capture, 3D digitization and printing, artificial intelligence and mechatronics – merging technological skill sets with critical discussions around artistry, humanity, and social possibilities.

Artists:

April 16 - Ben Baker, Assistant Professor of Philosophy at Colby College

April 30- Eto Otitigbe, Assistant Professor of Art at Brooklyn College

May 14- LaJuné McMillian, Multidisciplinary Artist and Educator

Join us from 5:00-7:00PM on any or all of the dates for these innovative programs:

Introduction: 5:00 - 5:10PM
Presentation: 5:10 - 5:30PM
Moderated Q&A: 5:30 - 5:50PM
Break: 5:50 - 6:00PM
Workshop: 6:00 - 7:00PM
Group Reflection: 7:00 - 7:30PM
Reception 7:30-8:30PM

Workshop Leaders: Thomas F. DeFrantz, Ted Quiballo, Craig Stevens, Michael A. Peshkin, Nick Marchuk, Darren Gergle

Series Convenors: Thomas F. DeFrantz (SLIPPAGE Lab/Performance Studies), Craig Stevens (Northwestern IT: Media & Technology Innovation/Anthropology), Ted Quiballo (Northwestern Libraries)

Sponsors: The Alice Kaplan Institute for the Humanities; The Alumnae of Northwestern University; Northwestern Libraries; Northwestern Depts. of IT and Performance Studies; and SLIPPAGE Lab.

Related Info

Register Now

Join the Buffett Institute for an discussion on the intersection of artificial intelligence (AI), law and society with Liane Huttner, Assistant Professor of Digital Law at Université Paris-Saclay. Moderated by Joanna Grisinger, Associate Professor of Instruction and Director of Legal Studies at Northwestern University, the conversation will explore the evolving legal landscape surrounding AI, its implications for governance, ethics and policy, and the challenges of regulating emerging technologies. An audience Q&A will follow the discussion.

Lunch will be provided beginning 12:15 p.m.

Please note that 720 University Place is not an ADA-accessible space. Increasing physical access to buildings and facilities is a goal of the University, but not all buildings and venues have been updated.

Register Now

(Re)visiting Foundation Models for Science and Beyond

Wei Wang, Leonard Kleinrock Professor in Computer Science and Computational Medicine, University of California, Los Angeles

Abstract: The emergence of large language models (LLMs) has introduced a new paradigm in data modeling. These models replace specialized models designed for individual tasks with unified models that are effective across a broad range of problems. In scientific domains, this shift not only transforms approaches to handling natural language tasks (e.g., scientific papers) but also suggests new methods for dealing with other data types (e.g., molecules, proteins, pathology images). In many fields, LLM has already shown great potential to accelerate scientific discovery. In this talk, I will present our recent work on LLMs, especially in the context of science and engineering research.

Related Info

Title: Control for Embodied Learning

Speaker: Todd Murphey, Northwestern University

Abstract: Embodied learning systems rely on motion synthesis to enable efficient and flexible learning during continuous online deployment. Motion motivated by learning needs can be found throughout natural systems, yet there is surprisingly little known about synthesizing motion to support learning for robotic systems. Moreover, robotic systems will need to collect data autonomously for learning, for instance when isolated for long period of time or when encountering novel environmental features. Learning goals create a distinct set of control-oriented challenges, including how to choose measures as objectives, synthesize real-time control based on these objectives, impose physics-oriented constraints on learning, and produce analyses that certify performance and safety with limited knowledge. This talk will discuss learning tasks that robots encounter, abstractions that enable regulating information content of observations, and recent progress on algorithms for generating action plans that facilitate learning.

Zoom: https://northwestern.zoom.us/j/96919208870

-----

To subscribe to the Applied Mathematics Colloquia List send a message to LISTSERV@LISTSERV.IT.NORTHWESTERN.EDU with the command:

add esam-seminar Youremail Firstname Lastname

In this session for PhDs and Postdocs, we’ll discuss the difference between academic CVs and resumes and best practices for writing compelling non-academic application documents. We’ll also explore how to leverage AI tools to analyze job descriptions to write more effective and targeted application materials.

Related Info

In recent years generative artificial intelligence (GAI) tools have increased in availability and popularity at universities, but not everyone knows how to use them for better library search results. In this 60-minute hybrid session we will demonstrate how to use Microsoft Copilot, a free GAI tool for all NU students and faculty, to improve searches with the gold standard of reliable content, library databases. This session is geared toward participants with little-to-no familiarity with generative artificial intelligence, and will also address the basics of what generative artificial intelligence is, how it works, its risks, and limitations. Students can learn about NU’s Copilot accounts here.

This workshop is presented by Tracy Coyne, Distance Learning and Professional Studies Librarian; Frank Sweis, User Experience Librarian; and Jeannette Moss, User Education Librarian.

A Northwestern Zoom Account is required to access this session.

Register Now

Abstract

Autonomous vehicles (AVs) rely on deep neural networks (DNNs) for critical tasks such as environment perception—identifying traffic signs, pedestrians, and lane markings—and executing control decisions like braking, acceleration, and lane changing. However, DNNs are vulnerable to adversarial attacks, including structured perturbations to inputs and misleading training samples that can degrade performance. This presentation begins with an overview of adversarial training, emphasizing the impact of input sizes on DNNs' vulnerability to cyberattacks. Subsequently, I will share our recent findings that explore the hypothesis that DNNs learn piecewise linear relationships between inputs and outputs. This conjecture is crucial for developing both adversarial attacks and defense strategies in machine learning security. The last part of the presentation will focus on recent work on using error-correcting codes to safeguard DNN-based classifiers.

Bio

Saif Jabari is a Global Network Associate Professor of Civil and Urban Engineering at New York University. His research interests center on theoretical aspects of traffic flow, specifically topics related to modeling uncertainty and emergent phenomena. The applications focus on traffic operations problems, including traffic state estimation and prediction, distributed traffic control, and cybersecurity.

Prior to joining NYUAD, Jabari was a Post-Doctoral Researcher in the Mathematical Sciences and Analytics Department at the IBM T.J. Watson Research Center in Yorktown Heights, NY. Jabari received his Ph.D. in Civil Engineering from the University of Minnesota, Twin Cities in 2012 and his B.Sc. degree in Civil Engineering from in the University of Jordan in 2001.

Related Info

Graph Neural Networks (GNNs) are a recent extension of the neural network machinery to the graph setting that resolve the challenge of extending deep learning methods the peculiarities of network data by convolving node features across neighbourhoods to embed nodes in Euclidean space. Heralded as the breakthrough for machine learning on graphs that would allow the same “AI renaissance” that standard neural networks have brought to Computer Vision and Natural Language Processing, GNNs have been suggested as a panacea for a wide number of tasks across disciplines. In the biological sciences alone, GNNs have been applied to molecular design, drug-drug interaction predictions, biological networks, knowledge graphs, and spatial transcriptomics.

Despite their popularity and widespread adoption, the theoretical foundations of GNNs remain underexplored. Fundamental questions about the mathematical principles driving their success, as well as their limitations, biases, and underlying statistical assumptions, are still unresolved. Notably, GNNs diverge significantly from traditional neural networks, with their architecture and function rooted in the unique properties of graph-structured data. These gaps in understanding could have significant implications in real-world applications, where issues revolving around bias and uncertainty must be rigorously addressed.

This workshop seeks to bring together researchers from statistics, computer science and computational biology to explore the theoretical and practical aspects of GNNs. The workshop will focus on topics revolving around three key themes:

• GNN Theory: Investigating foundational topics such as learning rates for classification and regression tasks, understanding the impact of different GNN architectures and the convolution operator.
• Uncertainty Quantification & Interpretability: Understanding the confidence and robustness of GNN predictions.
• Bias and Fairness: Exploring how GNNs may inadvertently propagate or amplify biases and ensuring equitable outcomes in their applications.

This two-day workshop will feature a series of focused deep-dive sessions, each dedicated to a core topic. These sessions will integrate expert talks with interactive discussions and structured brainstorming activities, led by small groups of participants, to foster collaboration and innovative thinking. The anticipated outcome of each session is the formulation of a precise research question and the initial framework of a research plan, providing participants with the opportunity to continue working on these topics beyond the workshop. The workshop's overarching goal is to produce a collaborative white paper that synthesizes the discussions, highlights key open questions, and outlines promising research directions in the theory and applications of Graph Neural Networks (GNNs).

Related Info

Register Now

Please join us for a casual fair to learn about grad school opportunities here at Northwestern!

Admissions teams and other program faculty and staff will be there to share information and answer questions. We will have materials to hand out and will be open to speaking one-on-one and looking at resumes. You are welcome to join whether you know which program interests you or you’d just like to learn more about the possibilities!
Lunch will be provided. We look forward to seeing you there!

The Master of Science programs featured:

- Master of Science in Biotechnology (MBP)

- Master of Science in Machine Learning and Data Science (MLDS)

- Master of Science in Artificial Intelligence (MSAI)

- Master of Science in Project Management (MPM)

- Master of Science in Robotics (MSR)

- Master of Science in Engineering Design Innovation (EDI)

- Master of Science in Theoretical and Applied Mechanics (TAM)

- Master of Science in Mechanical Engineering

- Master of Science in Material Science and Engineering

- Master of Science in Computer Engineering

- Master of Science in Computer Science

- Master of Science in Electrical Engineering

- Master of Science in Biomedical Engineering

- Master of Science in Chemical and Biological Engineering

- Master of Science in Energy and Sustainability (MSES)

- Master of Science in Law Program

Related Info