Research News

This research was led by Brandon Reagen, assistant professor of computer science and electrical and computer engineering, with Nandan Kumar Jha, a Ph.D. student under Reagen, and Zahra Ghodsi, who obtained her Ph.D. at NYU Tandon under Siddharth Garg, Institute associate professor of electrical and computer engineering.

Concerns surrounding data privacy are having an influence on how companies are changing the way they use and store users’ data. Additionally, lawmakers are passing legislation to improve users’ privacy rights. Deep learning is the core driver of many applications impacted by privacy concerns. It provides high utility in classifying, recommending, and interpreting user data to build user experiences and requires large amounts of private user data to do so. Private inference (PI) is a solution that simultaneously provides strong privacy guarantees while preserving the utility of neural networks to power applications.

Homomorphic data encryption, which allows inferences to be made directly on encrypted data, is a solution that addresses the rise of privacy concerns for personal, medical, military, government and other sensitive information. However, the primary challenge facing private inference is that computing on encrypted data levies an impractically high penalty on latency, stemming mostly from non-linear operators like ReLU (rectified linear activation function).

Solving this challenge requires new optimization methods that minimize network ReLU counts while preserving accuracy. One approach is minimizing the use of ReLU by eliminating uses of this function that do little to contribute to the accuracy of inferences.

“What we are to trying to do there is rethink how neural nets are designed in the first place,” said Reagen. “You can skip a lot of these time and computationally-expensive ReLU operations and still get high performing networks at 2 to 4 times faster run time.”

The team proposed DeepReDuce, a set of optimizations for the judicious removal of ReLUs to reduce private inference latency. The researchers tested this by dropping ReLUs from classic networks to significantly reduce inference latency while maintaining high accuracy.

The team found that, compared to the state-of-the-art for private inference DeepReDuce improved accuracy and reduced ReLU count by up to 3.5% (iso-ReLU count) and 3.5× (iso-accuracy), respectively.

The work extends an innovation, called CryptoNAS. Described in an earlier paper whose authors include Ghodsi and a third Ph.D. student, Akshaj Veldanda, CryptoNAS optimizes the use of ReLUs as one might rearrange how rocks are arranged in a stream to optimize the flow of water: it rebalances the distribution of ReLUS in the network and removes redundant ReLUs.

The investigators will present their work on DeepReDuce at the 2021 International Conference on Machine Learning (ICML) from July 18-24, 2021.

Julia Stoyanovich, director of the Center for Responsible AI (R/AI) at NYU Tandon, and assistant professor of computer science and engineering and of data science, co-authored this paper with Armanda Lewis, a graduate student pursuing her master’s at the NYU Center for Data Science.

The authors detail their development of and pedagogy for a technical course focused on responsible data science, which tackles the issues of ethics in AI, legal compliance, data quality, algorithmic fairness and diversity, transparency of data and algorithms, privacy, and data protection.

The ability to interpret machine-assisted decision-making is an important component of responsible data science that gives a good lens through which to see other responsible data science topics, including privacy and fairness. The researchers’ study includes best practices for teaching technical data science and AI courses that focus on interpretability, and tying responsible data science to current learning science and learning analytics research.

The work also explores the use of “nutritional labels” — a family of interpretability tools that are gaining popularity in responsible data science research and practice — for interpreting machine learning models.

• In the paper, the investigators offer a description of a unique course on responsible data science that is geared toward technical students, and incorporates topics from social science, ethics and law.
• The work connects theories and advances within the learning sciences to the teaching of responsible data science, specifically, interpretability — allowing humans to understand, trust and, if necessary, contest the computational process and its outcomes. The study asserts that interpretability is central to the critical study of the underlying computational elements of machine learning platforms.
• The collaborators assert that they are among the first to consider the pedagogical implications of responsible data science, creating parallels between cutting-edge data science research and cutting-edge educational research within the fields of learning sciences, artificial intelligence in education, and learning analytics and knowledge.

Additionally, the authors propose a set of pedagogical techniques for teaching the interpretability of data and models, positioning interpretability as a central integrative component of responsible data science.

This research was led by Monica Menendez, Global Network professor of civil and urban engineering, and Joseph Chow, deputy director of the C2SMART University Transportation Center at NYU Tandon.

This project proposes a flexible bus dispatching system using automated modular vehicle technology, and considers multimodal interactions and congestion propagation dynamics.

This study proposes a novel flexible bus dispatching system in which a fleet of fully automated modular bus units, together with conventional buses, serves the passenger demand. These modular bus units can either operate individually or combined (forming larger modular buses with a higher passenger capacity). This provides enormous flexibility to manage the service frequencies and vehicle allocation, reducing thereby the operating cost and improving passenger mobility.

The investigators developed an optimization model to determine the optimal composition of modular bus units and the optimal service frequency at which the buses (both conventional and modular) should be dispatched across each bus line. They explicitly accounted for the dynamics of traffic congestion and complex interactions between the modes at the network level, based on a recently proposed three-dimensional macroscopic fundamental diagram (3D-MFD). To the best of Chow and Menendez' knowledge, this is the first application of the 3D-MFD and modular bus units for the frequency setting problem in the domain of bus operations.

Using this system of analysis, the researchers were able to show improved costs across the system by adjusting the number of combined modular bus units and their dispatching frequencies to changes in car and bus passenger demand. A comparison with the commonly used approach that considers only the bus system (neglecting the complex multimodal interactions and congestion propagation) reveals the value of the proposed modeling framework.

This research was led by Joseph Chow, deputy director of the C2SMART University Transportation Center at NYU Tandon. Co-authors included Kaan Ozbay, Director, and Shri Iyer, Managing Director of C2SMART. Chow and Ozbay are professors in the department of Civil and Urban Engineering. 

The COVID-19 pandemic has affected travel behaviors and transportation system operations, and raised new challenges for public transit. Cities are grappling with what policies can be effective for a phased reopening shaped by social distancing.

The C2SMART researchers used a baseline model for pre-COVID conditions to create a new model representing travel behavior during the COVID-19 pandemic. They achieved this both by recalibrating the population agendas to include work-from-home, and by re-estimating the mode choice model (to fit observed traffic and transit ridership data) for the Center’s MATsim-NYC platform, a multi-agent simulation test bed for evaluating emerging transportation technologies and policies. They then analyzed the increase in car traffic due to the phased reopen plan guided by the state government of New York.

Analyzing four reopening phases and two reopening scenarios (with and without transit capacity restrictions), they found that a reopening with 100% transit capacity may only see as much as 73% of pre-COVID ridership and an increase in the number of car trips by as much as 142% of pre-pandemic levels. They also discovered that limiting transit capacity to 50% would decrease transit ridership further from 73% to 64% while increasing car trips to as much as 143% of pre-pandemic levels.

They noted that, while the increase appears small, the impact on consumer surplus is disproportionately large due to already increased traffic congestion. Many of the trips also get shifted to other modes like micromobility.

The findings imply that a transit capacity restriction policy during reopening needs to be accompanied by (1) support for micromobility modes, particularly in non-Manhattan boroughs, and (2) congestion alleviation policies that focus on reducing traffic in Manhattan, such as cordon-based pricing.

This research was led by Joseph Chow, deputy director of the C2SMART University Transportation Center at NYU Tandon and professor of civil and urban engineering, with researchers Brian Yueshuai He, Jinkai Zhu, Ziyi Ma, and Ding Wang.

Evaluation of the demand for emerging transportation technologies and policies can vary by time of day due to spillbacks on roadways, rescheduling of travelers’ activity patterns, and shifting to other modes that affect the level of congestion. These effects are not well-captured with static travel demand models.

Chow and his team calibrated and validated the first open-source multi-agent simulation model for New York City, called MATSim-NYC, to support agencies in evaluating policies such as congestion pricing. The simulation-based virtual test bed is loaded with a “synthetic” 2016 population of over eight million people, calibrated in a prior study. Model validation using transit stations and road links is comparable to NYPBM.

In a study published in Transport Policy the researchers used the model to evaluate a congestion pricing plan proposed by the Regional Plan Association, and found a much higher (127K) car trip reduction compared to the RPA report (59K). The team discovered that the Association’s pricing policy would impact the population segment making trips within Manhattan differently from the population segment of trips outside Manhattan: benefits from congestion reduction benefit the former by about 110%+ more than the latter.

The simulation can show that 37.3% of the Manhattan segment would be negatively impacted by the pricing compared to 39.9% of the non-Manhattan segment, which has implications for redistribution of congestion pricing revenues. The citywide travel consumer surplus decreases when the congestion pricing goes up from $9.18 to $14 both ways even as it increases for the Charging-related population segment. This implies that increasing pricing from $9.18 to $14 benefits Manhattanites at the expense of the rest of the city.

RPA congestion pricing policy would have net increase in consumer surplus. The results suggest toll revenue redistribution should focus on outer boroughs.

This research, under direction of Theodore (Ted) S. Rappaport, was led by graduate students Shihao Ju, Yunchou Xing, and Ojas Kanhere.

Driven by ubiquitous usage of mobile devices and the explosive growth and diversification of the Internet of Things (IoT), sixth-generation (6G) wireless systems will need to offer unprecedented high data rate and system throughput, which can be achieved in part by deploying systems transmitting and receiving at millimeter-wave (mmWave) and Terahertz (THz) frequencies (i.e., 30 GHz - 3 THz). These regions of the electromagnetic spectrum are capable of massive data throughput at near zero latency, key to future data traffic demand created by such wireless applications as augmented/virtual reality (AR/VR) and autonomous driving.

Importantly, the linchpin for successful deployment of mmWave and THz systems for 6G wireless communications will be their performance in indoor scenarios. Therefore, accurate THz channel characterization for indoor environments is essential to realizing the designs of transceivers, air interface, and protocols for 6G and beyond.

To this end, NYU WIRELESS has introduced NYUSIM 3.0, the latest version of its MATLAB®-based open-source mmWave and sub-THz statistical channel simulation software, enabling the indoor MIMO channel simulations for frequencies from 500 MHz to 150 GHz with RF bandwidth of 0 to 800 MHz. The new NYUSIM 3.0 is publicly available with a simple MIT-style open source acknowledgement license. To date, NYUSIM has been downloaded over 80,000 times.

NYUSIM 3.0 implemented a 3-D indoor statistical channel model for mmWave and sub-THz frequencies following the mathematical framework of the 3-D outdoor statistical channel model adopted in earlier versions of NYUSIM. The indoor 3-D statistical channel model for mmWave and sub-THz frequencies, was developed from extensive radio propagation measurements conducted in an office building at 28 GHz and 140 GHz in 2014 and 2019 — in both line-of-sight (LOS) and non-line-of-sight (NLOS) scenarios. The team carefully measured over 15,000 power delay profiles to study temporal and spatial channel statistics such as the number of time clusters, cluster delays, and lobe angular spreads.

The adopted channel models for Version 3.0 are elaborated in a upcoming paper, “Millimeter Wave and Sub-Terahertz Spatial Statistical Channel Model for an Indoor Office Building” (to appear in IEEE Journal on Selected Areas in Communications, Special Issue on Terahertz Communications and Networking in the second quarter 2021) by a team of three students at NYU WIRELESS, and the Department of Electrical and Computer Engineering under Rappaport’s guidance, led by Ph.D. student Shihao Ju. Besides proposing a unified indoor channel model across mmWave and sub-THz bands based on the team’s indoor channel measurements, the work provides a reference for future standards development above 100 GHz.

This research was conducted by Jin Kim Montclare, Professor of Chemical and Biomolecular Engineering; and former students Yao Wang, a recent Ph. D. graduate, and Xiaole Wang, a former M.S. student.

Protein hydrogels are versatile 3-dimensional macromolecular structures with an astonishing variety of potential applications, many of them in medicine, including tissue engineering and wound healing. Because of their hydrophilic properties and internal architecture, these compounds can even trap and deliver drugs directly to targets, opening up a host of potential applications involving safe delivery of cytotoxic compounds that are standard treatment for cancer and other diseases.

To have such “Swiss Army Knife” capabilities, these typically soft materials must be imbued with properties conferring static and dynamic mechanical strength that enables them to carry a molecular payload and know when to release it.

Taking up this challenge, Montclare and her former students built upon recent work developing a photo-crosslinkable triblock copolymer protein hydrogel called CEC-D, a compound with limited viscoelastic mechanical and moderate sustained release properties. In the new work they explored the potential of transition metal cations (positively charged ions) to enhance the mechanical properties of CED-D, including its ability to encapsulate and release the small molecule curcumin, known for its anti-inflammatory properties.

In the paper, “Effect of Divalent Metal Cations on the Conformation, Elastic Behavior, and Controlled Release of a Photocrosslinked Protein Engineered Hydrogel,” published in the ACS publication Applied Bio Materials, the investigators found that the hydrogels coordinated with divalent metal ions such as Zn2+, Cu2+, and Ni2+ demonstrated control over the encapsulation and release of curcumin, a discovery suggesting that cation-tuned hydrogels constitute a promising drug delivery platform with tunable physicochemical properties.

“Depending on the metal, we can control the structure, mechanical stiffness and small molecule delivery of the hydrogel,” said Montclare. “This has important implications for drug delivery and this knowledge can be used to tailor vehicles to deliver specific therapeutics. For example, we can tailor these materials to fabricate wound dressings that improve healing by triggering drug release in the presence of metals.”

Edward Wolf, professor emeritus at the NYU Tandon School of Engineering led this research

The 2020 hurricane season had a record breaking 30 named storms, 13 of which became hurricanes and six became major hurricanes. Compare that to the average storm season, which historically produces 12 named storms, six hurricanes, and three major hurricanes. It’s clear that hurricanes are increasing in both number and intensity.

Researchers have been studying the relation between water temperature and hurricane frequency for years now. There have been suggestions that as temperature increases, storms become more powerful.

Now, Tandon researcher Edward Wolf has released new research that confirms both facts. In studying recent hurricane data, Wolf was able to pinpoint the ocean temperature where it becomes possible for hurricanes to form. That temperature is approximately 26.5 degrees Celsius. Much like how water boils at 100 degrees Celsius (and not a degree lower), weather patterns cannot phase change into a hurricane until this water temperature is met.

Wolf also developed a simple algorithm that can predict the severity of a storm by measuring the temperature of the water beneath it. As the temperature goes up, the severity of the storm increases in a consistent and measurable manner. Not only does this prove that water temperature and storm strength are directly linked, it could be a tool to efficiently gauge the strength of a storm — an early warning system that could help communities in its path prepare.

Wolf’s research also provides one surprising detail: the algorithm describing how storm severity increases in proportion to ocean temperatures finds a direct analog in ferromagnetism — the strength of an iron magnet’s field: the temperature-defined phase change of ferromagnetism follows the same critical exponent formula T-Tc ^⅓ determining shift to magnetism at specific temperatures.

Researchers are now able to use the vast scientific literature on ferromagnetism in order to study hurricane formation, which by its nature has less raw data to work with. Wolf was able to use previous iron studies to fine tune his algorithm, producing even finer data.

This research was conducted by Jin Kim Montclare, Professor of Chemical and Biomolecular Engineering; and former students Yao Wang, a recent Ph. D. graduate, and Xiaole Wang, a former M.S. student.

Protein hydrogels, 3-dimensional macromolecular structures that do not dissolve in water (in spite of being hydrophilic), can hold large quantities of aqueous solutions due to the network formed from chemical or physical crosslinking. Partly because of this they have many medical applications including tissue engineering, wound healing and drug delivery.

These materials can be synthesized by crosslinking polymers chemically via covalent bonds, or physically via non-covalent interactions, or a mixture of both. One way of doing this is through photo-initiated crosslinking, wherein chemically inert groups become photo-reactive once exposed to certain wavelengths of light.

“The advantage of employing photochemical reactive groups over traditional chemical reagents is that they give the user spatiotemporal control over polymerization,“ said Montclare. “In other words, the biopolymers bearing the photocrosslinkers can be generated and crosslinked at some later step under various conditions to achieve control over encapsulation and release of therapeutic agents.”

In this research, the team designed a macromolecular triblock polymer comprising two different self-assembling domains derived from elastin protein (E) and coiled-coil protein (C), that can be photopolymerized due to an a NHS-diazirine (D) crosslinker to produce a CEC-D hydrogel.  

In the work, “Free-Standing Photocrosslinked Protein Polymer Hydrogels for Sustained Drug Release,” in Biomacromolecules, a publication of the American Chemical Society, the investigators determined the best photocrosslinker concentration and exposure time necessary to create this independent hydrogel.

The researchers found that, overall, CEC-D hydrogel exhibits comparable characteristics, including stability, drug release profile, and elastic behavior to other hydrogels.

Because it can be used for drug delivery with high encapsulation and a low but significant release of curcumin, CEC-D has been proven to be capable of a sustained release of a given drug over a week's time.

This research is led by Vikram Kapila, professor of mechanical and aerospace engineering. Principal authors are Ph.D. students Ashwin Rajkumar, Master's students Fabio Vulpi and Satish Reddy Bethi, and Preeti Raghavan of Johns Hopkins University School of Medicine and Rusk Rehabilitation at NYU School of Medicince.

Recent years have seen a brisk rise in development and deployment of digital health systems using such technologies as wearable sensors and embedded controllers to enhance access to medical diagnostics and treatments. Because of an accelerating trend in the number of stroke survivors requiring rehabilitation, healthcare services worldwide are considering technological solutions to enhance accessibility to assessment and treatment, particularly during the past year’s period of enforced quarantine due to COVID-19.

Some of the challenges faced by these technologies are clinical acceptance, high equipment cost, accuracy, and ease of use.

To address these limitations, the researchers designed wearable inertial sensors for exergames (WISE), a system that includes an animated virtual coach to deliver instruction, and a subject- model whose movements are animated by real-time sensor measurements from the WISE system worn by a subject. The paper examines the WISE system’s accuracy and usability for the assessment of upper limb range of motion (ROM).

The system uses five wearable sensor modules affixed to a user’s upper body: above the wrist on the left and right forearms, above the elbow on the left and right arms, and on the back. Each WISE sensor module consists of a sensor interfaced with a microcontroller soldered to a printed circuit board, which is connected to a lithium-ion battery. A storage and calibration cube is designed and 3D-printed to house the five WISE system modules and simultaneously calibrate all the sensors prior to placement on a subject. The microcontroller retrieves absolute orientation measurement from the sensor and wirelessly streams it to a computer. The investigators used a Unity3D-based exergame interface to animate the sensor data into a 3D-human model.

Seventeen neurologically intact subjects were recruited to participate in a usability study of the WISE system. The subjects performed a series of shoulder and elbow exercises for each arm instructed by the animated virtual coach; accuracy of the ROM measurements obtained with the WISE system were compared with those obtained with a system using the Microsoft Kinect markerless motion capture system (a platform used for exergames and often tested for rehabilitation capabilities). The results suggest the WISE system performs as well as Kinect.

The researchers plan future studies with patient populations in clinical and tele-rehabilitation settings.

This work is supported in part by the National Science Foundation DRK-12 Grant DRL-1417769, RET Site Grant EEC-1542286, and ITEST Grant DRL-1614085; NY Space Grant Consortium Grant 48240-7887; and Translation of Rehabilitation Engineering Advances and Technology (TREAT) grant NIH P2CHD086841.