Eight UMass Amherst Faculty Members Receive NSF CAREER Awards

Over the course of the 2023-24 academic year, eight faculty members across the UMass Amherst campus were named the recipients of five-year U.S. National Science Foundation (NSF) CAREER awards.<_o3a_p>

The Faculty Early Career Development (CAREER) Program is a foundation-wide activity that offers NSF awards in support of early career faculty who have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organization.<_o3a_p>

The College of Engineering was awarded four CAREER grants this year, bringing its total to 24 awards within the last five academic years. This year's awardees include Emily Kumpel (civil and environmental engineering), Robert Niffenegger (electrical and computer engineering), Govindarajan Srimathveeravalli (mechanical and industrial engineering) and Jinglei Ping (mechanical and industrial engineering).<_o3a_p>

The College of Natural Science (CNS) has been awarded two CAREER grants during this cycle, bringing its total to 63. This year's recipients include Annie Raymond (mathematics and statistics) and Varghese Mathai (physics).<_o3a_p>

Manning College of Information and Computer Sciences (CICS) professors Hui Guan and Negin Rahimi were awarded CAREER grants for their work on long-range wireless sensing and the development of search engines that work on a conversational model, respectively. The awards for Guan and Rahimi bring the cumulative number of CAREER awards for CICS to 38.<_o3a_p>

Hui Guan

Hui Guan (CICS) has been awarded $669,583 to help integrate two cutting-edge areas of computer science: edge computing and artificial intelligence-in this case, deep neural networks (DNNs). However, the limited resources on edge platforms, such as edge servers and Internet of Things devices, hinder the ability to deliver fast and accurate responses to queries from deep learning prediction tasks.<_o3a_p>

"Because these DNN models are currently designed to always be able to meet peak demand, they require too much computing power," says Guan. As a result, only some deep-learning tasks and smaller DNN models suitable for edge deployment are feasible. <_o3a_p>

To overcome this limitation, this project explores a new adaptive approach in building deep learning systems. The systems will make real-time adjustments to the DNNs executed for prediction tasks based on the varying resource demands arising from three critical dimensions: variable task complexity, fluctuating inference workloads, and resource contention in multi-tenant edge environments. The goal is to optimize both system efficiency and accuracy. Realizing the envisioned adaptiveness will facilitate the effective deployment of deep learning techniques across diverse applications and environments. <_o3a_p>

As Guan explains, there is currently a fundamental constraint preventing the implementation of deep neural network models on resource-constrained devices, such as wearable medical sensors, robots, or self-driving cars.
<_o3a_p>

Emily Kumpel

Emily Kumpel (civil and environmental engineering) has been awarded $549,834 to study contamination in mid-sized water storage tanks. <_o3a_p>

Wooden water tanks on New York City high rises are one example, "but internationally, they're ubiquitous," she says. "You go to any of the countries where piped water is only provided at intervals-which it is for 1 billion people in the world-and there are these tanks on every single roof." <_o3a_p>

Her research aims to understand what happens when these building-scale tanks are contaminated and evaluate the possible solutions. Keeping cost and water use low for these interventions is key because these are already water-scarce areas that cannot afford to flush 10,000 liters of water. She also plans to do fieldwork in Kenya to better understand how people use home water storage systems and how engineers can reassess the design to maximize reliability.

Finally, her research will serve as the basis for high school students to build their own scale model tanks and understand the hydraulics of how contaminated water endures. For the general public, she will create a map to monitor water outages caused by natural disasters, drought or infrastructure issues as a way to understand how water outages are occurring and how they affect people.<_o3a_p>

Varghese Mathai

Varghese Mathai (physics) has been awarded $554,191 to investigate fluid-structure interactions where the interacting structure is thin and ultrasoft (like Jell-O), and where the flow is turbulent. The complexity of this topic is increased by the fact that such thin elastic materials can undergo large, nonlinear, flow-induced shape morphings, and there is a knowledge gap in our understanding of the behavior of such ultrasoft solids in turbulent flows.<_o3a_p>

The principal aim of this project is to develop a deeper understanding of this special combination where nonlinear elasticity and (nonlinear) fluid dynamics are entangled, leading to the emergence of new flow properties. <_o3a_p>

"Studying the fundamental mechanics of ultra-soft materials within turbulent flows can aid in the development of new technologies for energy extraction and propulsion," says Mathai. "For example, these soft materials can be designed to shape-morph and adapt to varying tidal currents, improving the energy-extraction efficiency."<_o3a_p>

The project will furthermore seek to develop and systematically explore flapping membrane hydrofoils in a water flume facility using a three-degree-of-freedom platform, study the implications of elastic shape-morphing on unsteady lift and flow-induced resonance phenomena, and understand the mechanisms by which the membrane oscillations can be tuned to control turbulence and modulate drag.<_o3a_p>

This approach is expected to yield fundamental insight into the mechanics of ultrasoft materials in turbulent flow environments and provide opportunities for future innovation in tidal and fluvial energy extraction. Furthermore, Mathai will look to develop a class of soft materials for use in flow control, drag modulation and energy extraction applications.<_o3a_p>

Robert Niffenegger

Robert Niffenegger (electrical and computer engineering) has been awarded $624,196 to create a qubit system-on-a-chip by combining integrated ion trap chip technology with integrated photonics. The miniature quantum system aims to improve the performance, portability and scalability of quantum technologies by leveraging integration. <_o3a_p>

In a trapped-ion quantum computer, a laser precisely controls the atomic states of the ion, making them qubits, the quantum equivalent of a computer bit. However, adding more qubits to quantum computers without compromising performance has been a long-standing challenge. <_o3a_p>

"To compete with a classical supercomputer, a handful of qubits is not enough, and, if we want to deliver laser light to one million trapped ion qubits, it only scales if the chip is aligning the laser beams for you," as opposed to manual alignment to each qubit, says Niffenegger.<_o3a_p>

With this in mind, the goal is to create a quantum system-on-a-chip where integrated photonics route the laser light through the chip to the ion qubits, with a new challenge being to integrate the entire laser system into the same chip as the ion trap. The resulting quantum chip will be 1,000 times smaller, enabling the development of portable optical clocks and portable quantum computers. <_o3a_p>

"Leveraging integrated technology for the optical infrastructure required for quantum computers could make them small enough to fit in your pocket, replacing entire laboratories of delicate optics and lasers," he says. <_o3a_p>

Niffenegger's lab has already shown the first qubit operations with integrated laser sources but on separate chips for now. They're already working on the next steps with their collaborators at U.C. Santa Barbara, co-designing a new chip with the entire quantum system on a single chip for the first time.<_o3a_p>

Jinglei Ping

Jinglei Ping (mechanical and industrial engineering) has been awarded $550,000 to develop a new way to detect the presence and concentration of multiple types of DNA or RNA in a sample at high sensitivities. He previously demonstrated a new method to identify a single genetic marker with greater speed and sensitivity than traditional testing methods by putting samples into an alternating electric field, which makes the genetic material "dance."<_o3a_p>

Because each molecule has a signature movement pattern, researchers can easily identify the presence and concentration of the genetic material in question. However, this kind of testing is too limited for more complex diseases. For instance, researchers may identify that a blood sample has abnormally high levels of one particular type of micro-RNA.<_o3a_p>

"There are hundreds of micro-RNA in human blood, and if we just pick one out, that does not provide much information," he says. "There could be thousands of problems related to that concentration variation."<_o3a_p>

Ping will build on this method by adding the ability to distinguish between multiple genetic strands within a sample.<_o3a_p>

"We can combine the concentration information with machine learning to narrow down the possibilities of what the disease is," he says. "This is a promising pathway for point-of-care diagnosis for complex diseases like cancers and Parkinson's disease."<_o3a_p>

The final aim of the project is a compact, user-friendly device that can offer point-of-care disease diagnoses within minutes instead of days or weeks.<_o3a_p>

Negin Rahimi

Negin Rahimi (CICS) has been awarded $599,916 to enable users of generative AI models to obtain an interpretable, diverse and unbiased set of alternative answers, viewpoints, subtopics or aspects as required for various questions or tasks in information access, where each distinct answer or viewpoint is faithfully attributable to a set of evidence and supporting information sources.<_o3a_p>

This project aims to make information access easier, more effective and more trustworthy for users. Given that search is among the most common online activities, this project is positioned to have a substantial impact on society, promoting a more comprehensive understanding of topics, encouraging critical thinking, and facilitating informed decision-making.<_o3a_p>

Large generative AI models, such as ChatGPT, are widely used for information seeking purposes. Compared to traditional search engines, they provide a coherent narrative, which could potentially facilitate the exploratory phase of users' searches. Generative AI responses are more readable, coherent and contextually appropriate; hence, they sound authoritative and definitive. However, existing generative AI models are subject to problems such as hallucinations, unsupported misleading answers, outright misinformation and hidden biases. Another issue is that the majority of user queries are ambiguous. Current systems, including those that employ generative AI models, do not appropriately consider ambiguity by providing users with alternative answers to their queries.<_o3a_p>

This project proposes the development of novel retrieval models to enhance the relevance, diversity and interpretability of their results. This project will develop models for multi-granular diversification of search results to significantly improve the generalizability of retrieval models in providing diverse results for open-domain queries. In addition, this project enables the full utilization of search results by AI systems through explanations of their relevance and diversity. Building on top of explainable search results, the project introduces explanation-based optimization of search results. This involves improving search results based on reasoning over failures of retrieval models.<_o3a_p>

"With techniques like these," says Rahimi, "we can improve the accuracy, diversity, and trustworthiness of information from LLM-based information access systems. And with more diverse results, these systems promote a more comprehensive understanding of topics, encourage critical thinking, and facilitate informed decision making."<_o3a_p>

Annie Raymond

Annie Raymond (mathematics and statistics) has been awarded $450,000 to further the understanding of "graph profiles" - objects that record all possible relationships between local patterns. Many problems in engineering, science, economics, and social sciences involve complicated systems that can be represented as graphs.<_o3a_p>

"For example, the brain can be viewed as a graph where there are over 100 billion nodes corresponding to the neurons and where the connections represent synapses," says Raymond.<_o3a_p>

Computing different properties of these graphs yields valuable information about the original problems, but it is difficult to do so because of the size of the graphs. One technique to study such large graphs is to understand them locally by determining how prevalent certain small substructures are, for example through homomorphism densities.

The research component of this project will focus on four objectives: To compute graph profiles, including some in more than two dimensions; To study the strengths and limitations of different techniques (e.g., (rational) sums of squares, sums of nonnegative circuits) in proving inequalities over graph profiles; To better understand for which classes of inequalities certification over graph profiles is (un)decidable, and; To build theory and compute tropicalizations of graph profiles, which are simpler and yet capture all valid pure binomial inequalities, and to use these computations to resolve problems in extremal graph theory.<_o3a_p>

Govindarajan Srimathveeravalli

Govindarajan Srimathveeravalli (mechanical and industrial engineering) has been awarded $558,436 to improve the deliverability of drugs locally using microseconds-long electric pulses on endothelial cells. These cells line the inside of blood vessels and decide what does or doesn't get into the nearby organs. However, in disease states, these cells can either become too leaky or not leaky enough.<_o3a_p>

"These endothelial gatekeeper cells are one of the barriers for making sure that the right drug gets to the right cell at the right concentration," he says. "If the right drug doesn't get to the cancer cell, for example, in sufficiently high enough concentrations, it's not going to work."<_o3a_p>

He aims to target the blood vessels where a higher concentration of a drug is needed with short bursts of electricity, which triggers a transient change in the cell's scaffolding that creates gaps. These gaps allow more of a drug to penetrate into the target location.<_o3a_p>

In patients with advanced cancers involving inoperable tumors, this technology offers a chance to deliver high doses of drugs to shrink tumors to an operable size, resulting in better patient outcomes.<_o3a_p>

Public link

Please use the above public link if you want to share this noodl on another website.