Title
Topic
-
Grant from National Institutes of Health to combat drug-resistant pathogens
The project, titled “Repurposing Gram-Positive Antibiotics for Gram-Negative Bacteria Using Antibiotic Adjuvants,” studies “The multidrug-resistant (MDR) sepsis pathogen Acinetobacter baumanni,” writes professor of biology Edward Geisinger. “Current treatment options for infections with these bacteria are extremely limited. Our research examines a class of small molecules called antibiotic adjuvants that greatly boost the activity of several existing antibiotics against A. baumanniim, with the goal of developing new combination approaches to treat MDR infections.”
-
Making AI more secure with privacy-preserving machine learning
“Electrical and computer engineering assistant professor Xiaolin Xu, in collaboration with Wujie Wen from Lehigh University and Caiwen Ding from the University of Connecticut, was awarded a $1.2M NSF grant for ‘Accelerating Privacy-Preserving Machine Learning as a Service: From Algorithm to Hardware.'”
-
NSF CAREER Award to protect AI-enabled systems from attack
“Electrical and computer engineering assistant professor Xiaolin Xu was awarded a $600,000 NSF CAREER Award for ‘Securing Reconfigurable Hardware Accelerator for Machine Learning: Threats and Defenses.'”
-
Securing scientific cyberinfrastructures from advanced attacks
“Electrical and computer engineering assistant professor Xiaolin Xu is leading a $1.2 million NSF grant, in collaboration with professor of electrical and computer engineering Miriam Leeser and Mike Zink from the University of Massachusetts, for ‘CAREFREE: Cloud infrAstructure ResiliencE of the Future foR tEstbeds, accelerators and nEtworks.'”
-
‘Flourishing in the Everyday: Moving Beyond Damage-Centered Design in HCI for BIPOC Communities’
“Research and design in human-computer interaction centers problem-solving, causing a downstream effect of framing work with and for marginalized communities predominantly from the lens of deficit and damage. … However, we observe an additional need to center positive aspects of humanity, such as joy, pleasure, rest, and cultural heritage, particularly for Black, Indigenous, and People of Color. In this paper, we present three case studies of existing technologies that center BIPOC flourishing to provide an alternative path for HCI.” Find the paper and the full list of authors in the 2023 ACM Designing Interactive Systems Conference proceedings.
-
‘That’s a Tough Call: Studying the Challenges of Call Graph Construction for WebAssembly’
“WebAssembly is a low-level bytecode format that powers applications and libraries running in browsers, on the server side, and in standalone runtimes. Call graphs are at the core of many interprocedural static analysis and optimization techniques. However, WebAssembly poses some unique challenges for static call graph construction. … This paper presents the first systematic study of WebAssembly-specific challenges for static call graph construction and of the state-of-the-art in call graph analysis.” Find the paper and the full list of authors in the Proceedings of the 32nd ACM SIGSOFT International Symposium on Software Testing and Analysis.
-
‘Systematic Comparisons Between Lyme Disease and Post-Treatment Lyme Disease Syndrome in the U.S. With Administrative Claims Data’
“Post-treatment Lyme disease syndrome (PTLDS) is used to describe Lyme disease patients who have the infection cleared by antibiotic but then experienced persisting symptoms of pain, fatigue, or cognitive impairment. Currently, little is known about the cause or epidemiology of PTLDS. … We conducted a data-driven study with a large nationwide administrative dataset, which consists of more than 98 billion billing and 1.4 billion prescription records between 2008 and 2016, to identify unique aspects of PTLDS that could have diagnostic and etiologic values.” Find the paper and the full list of authors at EBioMedicine.
-
‘New Sampling Lower Bounds via the Separator’
“Suppose that a target distribution can be approximately sampled by a low-depth decision tree, or more generally by an efficient cell-probe algorithm. It is shown to be possible to restrict the input to the sampler so that its output distribution is still not too far from the target distribution, and at the same time many output coordinates are almost pairwise independent. This new tool is then used to obtain several new sampling lower bounds and separations, including a separation between AC0 and low-depth decision trees, and a hierarchy theorem for sampling.”
-
‘On Correlation Bounds Against Polynomials’
“We study the fundamental challenge of exhibiting explicit functions that have small correlation with low-degree polynomials over 𝔽₂. Our main contributions include: …2) We propose a new approach for proving correlation bounds with the central ‘mod functions.’ …3) We prove our conjecture for quadratic polynomials. … We express correlation in terms of directional derivatives and analyze it by slowly restricting the direction.4) We make partial progress on the conjecture for cubic polynomials, in particular proving tight correlation bounds for cubic polynomials whose degree-3 part is symmetric.” Find the paper and full list of authors at the Dagstuhl Research Online Publication Server.
-
‘Integrating Symmetry into Differentiable Planning with Steerable Convolutions’
“In this paper, we study a principled approach on incorporating group symmetry into end-to-end differentiable planning algorithms and explore the benefits of symmetry in planning. To achieve this, we draw inspiration from equivariant convolution networks and model the path planning problem as a set of signals over grids.” Find the paper and the full list of authors at Open Review.
-
‘Symmetries, Flat Minima and the Conserved Quantities of Gradient Flow’
“Empirical studies of the loss landscape of deep networks have revealed that many local minima are connected through low-loss valleys. Yet, little is known about the theoretical origin of such valleys. We present a general framework for finding continuous symmetries in the parameter space, which carve out low-loss valleys. Our framework uses equivariances of the activation functions and can be applied to different layer architectures.” Find the paper and the full list of authors at Open Review.
-
‘The Surprising Effectiveness of Equivariant Models in Domains With Latent Symmetry’
“Extensive work has demonstrated that equivariant neural networks can significantly improve sample efficiency and generalization by enforcing an inductive bias in the network architecture. These applications typically assume that the domain symmetry is fully described by explicit transformations of the model inputs and outputs. However, many real-life applications contain only latent or partial symmetries which cannot be easily described by simple transformations of the input. … We discover, surprisingly, that imposing equivariance constraints that do not exactly match the domain symmetry is very helpful in learning the true symmetry.” Find the paper and full list of authors at Open Review.
-
‘Equivariant Single View Pose Prediction via Induced and Restricted Representations’
“Learning about the three-dimensional world from two-dimensional images is a fundamental problem in computer vision. An ideal neural network architecture for such tasks would leverage the fact that objects can be rotated and translated in three dimensions to make predictions about novel images. However, imposing SO(3)-equivariance on two-dimensional inputs is difficult because the group of three-dimensional rotations does not have a natural action on the two-dimensional plane. … We show that an algorithm that learns a three-dimensional representation of the world from two dimensional images must satisfy certain geometric consistency properties.” Find the paper and full list of authors at ArXiv.
-
‘Location-Independent GNSS Relay Attacks: A Lazy Attacker’s Guide to Bypassing Navigation Message Authentication’
“In this work, we demonstrate the possibility of spoofing a GNSS receiver to arbitrary locations without modifying the navigation messages. … Prior work required an adversary to record the GNSS signals at the intended spoofed location and relay them to the victim receiver. Our attack demonstrates the ability of an adversary to receive signals close to the victim receiver and in real-time generate spoofing signals for an arbitrary location without modifying the navigation message contents.” Find the paper and full list of authors in the 16th ACM Conference on Security and Privacy in Wireless and Mobile Networks proceedings.
-
‘UE Security Reloaded: Developing a 5G Standalone User-Side Security Testing Framework’
“Security flaws and vulnerabilities in cellular networks lead to severe security threats given the data-plane services that are involved, from calls to messaging and Internet access. While the 5G Standalone (SA) system is currently being deployed worldwide, practical security testing of User Equipment (UE) has only been conducted and reported publicly for 4G/LTE and earlier network generations. In this paper, we develop and present the first open-source based security testing framework for 5G SA User Equipment.” Find the paper and the full list of authors in the 16th ACM Conference on Security and Privacy in Wireless and Mobile Networks proceedings.
-
‘Encrypted Databases Made Secure Yet Maintainable’
“State-of-the-art encrypted databases (EDBs) can be divided into two types: one that protects the whole DBMS engine in a trusted domain, and one that protects only operators that support queries over encrypted data. Both types have limitations when dealing with malicious database administrators (DBAs). The first type either exposes the data to DBAs or makes maintenance operations difficult if the DBA role is eliminated. The second type is vulnerable to abuse of the operator interfaces; … we devise a smuggle attack that enables DBAs to secretly and effectively access data.” Find the paper and full list of authors at USENIX.
-
‘Boosting Multitask Learning on Graphs Through Higher-Order Task Affinities’
“Predicting node labels on a given graph is a widely studied problem with many applications. … This paper considers predicting multiple node labeling functions on graphs simultaneously and revisits this problem from a multitask learning perspective. … Due to complex overlapping patterns, we find that negative transfer is prevalent when we apply naive multitask learning to multiple community detection, as task relationships are highly nonlinear across different node labeling. To address the challenge, we develop an algorithm to cluster tasks into groups based on a higher-order task affinity measure.” Find the paper and full list of authors at ArXiv.
-
‘Toward Computationally-Supported Roleplaying for Perspective-Taking’
“Designing and studying computationally-supported roleplaying for changing social perspectives of players is a complex and challenging problem. As indispensable components of roleplaying games (RPGs), narratives have the potential to promote successful perspective-taking. … We first present the design of a visual novel style RPG scenario addressing xenophobia and bullying, using an interactive narrative powered by a computational narrative engine. We then report on a usability evaluation of our interactive narrative system and an empirical evaluation of the RPG’s effectiveness in promoting successful perspective-taking through a crowdsourced online experiment.” Find the full list of authors in the International Conference on Human-Computer…
-
‘Summarizing, Simplifying and Synthesizing Medical Evidence Using GPT-3 (With Varying Success)’
“Large language models, particularly GPT-3, are able to produce high quality summaries of general domain news articles in few- and zero-shot settings. However, it is unclear if such models are similarly capable in more specialized, high-stakes domains such as biomedicine. In this paper, we enlist domain experts (individuals with medical training) to evaluate summaries of biomedical articles generated by GPT-3, given zero supervision. … We design an annotation scheme for evaluating model outputs, with an emphasis on assessing the factual accuracy of generated summaries.” Find the paper and the full list of authors at ArXiv.
-
‘Multilingual Simplification of Medical Texts’
“Automated text simplification aims to produce simple versions of complex texts. This task is especially useful in the medical domain, where the latest medical findings are typically communicated via complex and technical articles. This creates barriers for laypeople seeking access to up-to-date medical findings, consequently impeding progress on health literacy. … This work addresses this limitation via multilingual simplification, i.e., directly simplifying complex texts into simplified texts in multiple languages. We introduce MultiCochrane, the first sentence-aligned multilingual text simplification dataset for the medical domain in four languages: English, Spanish, French, and Farsi.” Find the paper and full list of authors…
-
‘Appraising the Potential Uses and Harms of LLMs for Medical Systematic Reviews’
“Medical systematic reviews are crucial for informing clinical decision making and healthcare policy. But producing such reviews is onerous and time-consuming. Thus, high-quality evidence synopses are not available for many questions and may be outdated even when they are available. Large language models (LLMs) are now capable of generating long-form texts, suggesting the tantalizing possibility of automatically generating literature reviews on demand. However, LLMs sometimes generate inaccurate (and potentially misleading) texts by hallucinating or omitting important information. … [Here], we seek to qualitatively characterize the potential utility and risks of LLMs.” Find the paper and full list of authors at ArXiv.
-
‘USB: A Unified Summarization Benchmark Across Tasks and Domains’
“An abundance of datasets exist for training and evaluating models on the task of summary generation. However, these datasets are often derived heuristically, and lack sufficient annotations to support research into all aspects of summarization. … We introduce a benchmark comprising 8 tasks that require multi-dimensional understanding of summarization, e.g., surfacing evidence for a summary, assessing its correctness, and gauging its relevance to different topics. We compare various methods on this benchmark and discover that on multiple tasks, moderately-sized fine-tuned models consistently outperform much larger few-shot prompted language models.” Find the paper and the full list of authors at ArXiv.
-
‘Automated Metrics for Medical Multi-Document Summarization Disagree with Human Evaluations’
“Evaluating multi-document summarization (MDS) quality is difficult. This is especially true in the case of MDS for biomedical literature reviews, where models must synthesize contradicting evidence reported across different documents. … We introduce a dataset of human-assessed summary quality facets and pairwise preferences to encourage and support the development of better automated evaluation methods for literature review MDS. We take advantage of community submissions to the Multi-document Summarization for Literature Review (MSLR) shared task to compile a diverse and representative sample of generated summaries.” Find the paper and the full list of authors at ArXiv.
-
‘Evaluating the Zero-Shot Robustness of Instruction-tuned Language Models’
“Instruction fine-tuning has recently emerged as a promising approach for improving the zero-shot capabilities of Large Language Models (LLMs) on new tasks. This technique has shown particular strength in improving the performance of modestly sized LLMs, sometimes inducing performance competitive with much larger model variants. In this paper we ask two questions: (1) How sensitive are instruction-tuned models to the particular phrasings of instructions, and, (2) How can we make them more robust to such natural language variation?” Find the paper and the full list of authors at ArXiv.
-
‘On Robot Grasp Learning Using Equivariant Models’
“Real-world grasp detection is challenging due to the stochasticity in grasp dynamics and the noise in hardware. Ideally, the system would adapt to the real world by training directly on physical systems. However, this is generally difficult due to the large amount of training data required by most grasp learning models. In this paper, we note that the planar grasp function is $\SE(2)$-equivariant and demonstrate that this structure can be used to constrain the neural network used during learning.” Find the paper and the full list of authors at ArXiv.
-
‘Probabilistic Symmetry for Multi-Agent Dynamics’
“Learning multi-agent dynamics is a core AI problem with broad applications in robotics and autonomous driving. While most existing works focus on deterministic prediction, producing probabilistic forecasts to quantify uncertainty and assess risks is critical for downstream decision-making tasks. … By leveraging symmetry, specifically rotation equivariance, we can improve not only the prediction accuracy but also uncertainty calibration. We introduce Energy Score, a proper scoring rule, to evaluate probabilistic predictions. We propose a novel deep dynamics model, Probabilistic Equivariant Continuous COnvolution (PECCO) for probabilistic prediction of multi-agent trajectories.” Find the paper and full list of authors in Proceedings of Machine Learning…