Accomplishments and achievements from the Visualization Imaging and Data Analysis (VIDA) Center

Work That Stands the Test of Time

Visual Exploration of Big Spatio-Temporal Urban Data: A Study of New York City Taxi Trips —  authored by Claudio Silva and Julian Freire, co-Directors of VIDA, among others — was named the IEEE VIS Test of Time Award at the VIS 2023 conference this year. The award is an accolade given to recognize articles published at previous conferences whose contents are still vibrant and useful today and have had a major impact and influence within and beyond the visualization community.

The decisions are based on objective measures such as the numbers of citations, and more subjective ones such as the quality and longevity and influence of ideas, outreach, uptake and effect not only in the research community, but also within application domains and visualization practice.

The paper has been officially cited over 600 times, and has led to more than 50 follow-on papers building on the initial research. With a unique data set and a revolutionary approach, they created — for the first time — a large, complex, dynamic urban data set that was leveraged for visualization in an urban context.

The paper was written as complicated data sets about urban living were becoming more-and-more available. It focused on a particularly important urban data set: taxi trips. Taxis are valuable sensors, and information associated with taxi trips can provide unprecedented insight into many different aspects of city life, from economic activity and human behavior to mobility patterns. But the data is complex, containing geographical and temporal components in addition to multiple variables associated with each trip. This problem was compounded due to the size of the data — there were on average 500,000 taxi trips each day in NYC at that time. 

The researchers proposed a model that allows users to visually query taxi trips. Besides standard analytics queries, it supported origin-destination queries that enable the study of mobility across the city. It also had the flexibility to allow users to explore and compare results. Their system implemented this model which supported interactive response times; made use of an adaptive level-of-detail rendering strategy to generate clutter-free visualization for large results; and showed hidden details to the users in a summary through the use of overlay heat maps. 

IEEE cited the paper for its “interactive hierarchical dimension ordering, spacing and filtering for exploration of high dimensional datasets”  which produced “thoughtful, elegant and powerful approach to managing the complexities of high dimensional data and reducing clutter in HighD visualizations.”

“The paper shows us how we can solve a problem through interactive visualization design, and presents some convincing options for future analysts and designers. These ideas underpin subsequent research on synthesizing new summary dimensions, contribute to contemporary thinking on explainability and have influenced the design of many other high dimensional visualization tools and techniques.”

Beyond Silva and Freire, the other authors are now tenured faculty members at other universities:

• Huy T. Vo at The City University of New York has published work in recent years on topics like predicting Taxi & Uber Demand.
• Jorge Poco, who is at the School of Applied Mathematics, Fundação Getulio Vargas in Brazil, who is working on visualizing areas including legal precedent, crime patterns, and spatio-temporal networks in communities.
• Nivian Ferreira, Universidade Federal de Pernambuco, Brazil who conducts research in areas including spatiotemporal visualizations of 3D urban analytics and navigating cities in virtual reality

At Tandon, the paper laid the groundwork for varying applications, including visualizing subway maps, bus transport patterns, rideshare prediction (as noted above), Urbane, which allowed for first-of-its kind spatio-temporal 3D city map visualization, accessibility via sidewalk mapping and open-source software to allow others to do the same to inform city planning/government.

More Award-Winning Researchers

While IEEE gives the Test of Time Award in recognition of the fact that “to improve the future, we must reflect on our past,” at NYU Tandon, VIDA researchers are also busy envisioning the future of their field, and the work being conducted currently provides a window into what that future might look like.. 

Guido Gerig, Institute Professor of Computer Science and Engineering and Biomedical Engineering as well as a member of the Visualization Imaging and Data Analysis Center (VIDA),  has been honored with the MICCAI Enduring Impact Award (EIA). The Enduring Impact Award is a prestigious annual prize awarded since 2009 to senior researchers whose work has made an enduring impact on the field of medical image computing and computer-assisted interventions. The award is granted by the Medical Image Computing and Computer-Assisted Intervention Society (MICCAI) based on a researcher's originality, successful clinical applications, publications, conferences, and education activities.

Data-driven systems employ algorithms to aid human judgment in critical domains like hiring and employment, school and college admissions, credit and lending, and college ranking. Because of their impacts on individuals, population groups, institutions, and society at large, it is critical to incorporate fairness, accountability, and transparency considerations into the design, validation, and use of these systems. 

Assistant Professor Julia Stoyanovich has garnered a $400,000 grant from the National Science Foundation (NSF) to investigate the core technical challenges inherent in the responsible design and validation of algorithmic rankers.

Stoyanovich is also the recipient of another recent NSF grant, this one to support her research into the role of HR Specialists and ways to empower them with the agency to reason about, validate, audit, and influence the AI-assisted hiring process.

Assistant Professor Rumi Chunara is focused on mitigating bias in healthcare, and she recently won a  data, human and planetary health grant from the Max Planck Society to discover ways of leveraging inclusive data and causal methods for doing so.

Congratulations are also due to Ph.D. candidate Lucas Rosenblatt on his NSF Graduate Research Fellowship, which will allow him to further explore open questions on data privacy, algorithmic fairness and AI safety, with an eye towards improving society and doing social good.

It’s A Snap

ITK-SNAP is a free, open-source, multi-platform software application used to segment structures in 3D and 4D biomedical images. It was originally developed by a team of students led by Institute Professor Guido Gerig, who envisioned a tool that would be easy to learn, with a limited feature set centered specifically on the task of image segmentation.

ITK-SNAP is nearing major milestones: as of this fall, the tool has been downloaded almost three-quarters of a million times, and the associated paper has been cited by almost 8,000 researchers.

The Paper Chase

Papers written recently by VIDA faculty members and students have been presented at some of the most prestigious data visualization events in the world, including IEEE VIS 2023; the Very Large Data Base (VLDB) Conference; International Conference on Acoustics, Speech, & Signal Processing (ICASSP); AutoML Conference; Information Processing in Medical Imaging (IPMI) ‘23; and the International Conference on Computer Vision (ICCV).

Among the highlights were:

• “Epistemic Parity: Reproducibility as an Evaluation Metric for Differential Privacy” was presented at the 2023 VLDB Conference. (The team includes three students from Ukraine, working under the auspices of a VIDA fellowship program.) At that same conference, a demo of “ERICA: Query Refinement for Diversity Constraint Satisfaction” by Stoyanovich and several international collaborators was given an honorable mention. 
• “ARGUS: Visualization of AI-Assisted Task Guidance in AR” garnered a Best Paper Honorable Mention Award at IEEE VIS 2023.
• “Robust Online Multiband Drift Estimation in Electrophysiology Data” landed in the top 3% to be a best student paper finalist at ICASSP.
• “AlphaD3M: An Open-Source AutoML Library for Multiple ML Tasks” was presented at the 2023 AutoML Conference and can be tried out at github.com/VIDA-NYU/alpha-automl.
• “E(3) × SO(3)-Equivariant Networks for Spherical Deconvolution in Diffusion MRI” was both presented at MIDL ‘23 and selected as one of eight best papers for a special issue of MedIA Journal.

VIDA By The Numbers

In just the last several months, VIDA researchers have

• Published 25 new papers–5 of them the recipients of best-paper honors
• Won two new NSF grants
• Welcomed 18 new members
• Grown to 47 Ph.D. students, 14 research engineers and post-doctoral researchers, and nine dedicated faculty members