publications

2025

1. CCS

   Busting the Paper Ballot: Voting Meets Adversarial Machine Learning

   Kaleel Mahmood, Caleb Manicke, Ethan Rathbun, and 5 more authors

   Jun 2025

   Abs arXiv Bib

   We show the security risk associated with using machine learning classifiers in United States election tabulators. The central classification task in election tabulation is deciding whether a mark does or does not appear on a bubble associated to an alternative in a contest on the ballot. Barretto et al. (E-Vote-ID 2021) reported that convolutional neural networks are a viable option in this field, as they outperform simple feature-based classifiers. Our contributions to election security can be divided into four parts. To demonstrate and analyze the hypothetical vulnerability of machine learning models on election tabulators, we first introduce four new ballot datasets. Second, we train and test a variety of different models on our new datasets. These models include support vector machines, convolutional neural networks (a basic CNN, VGG and ResNet), and vision transformers (Twins and CaiT). Third, using our new datasets and trained models, we demonstrate that traditional white box attacks are ineffective in the voting domain due to gradient masking. Our analyses further reveal that gradient masking is a product of numerical instability. We use a modified difference of logits ratio loss to overcome this issue (Croce and Hein, ICML 2020). Fourth, in the physical world, we conduct attacks with the adversarial examples generated using our new methods. In traditional adversarial machine learning, a high (50% or greater) attack success rate is ideal. However, for certain elections, even a 5% attack success rate can flip the outcome of a race. We show such an impact is possible in the physical domain. We thoroughly discuss attack realism, and the challenges and practicality associated with printing and scanning ballot adversarial examples.

   @article{Mahmood2025,
     title = {Busting the Paper Ballot: Voting Meets Adversarial Machine Learning},
     author = {Mahmood, Kaleel and Manicke, Caleb and Rathbun, Ethan and Verma, Aayushi and Ahmad, Sohaib and Stamatakis, Nicholas and Michel, Laurent and Fuller, Benjamin},
     year = {2025},
     month = jun,
     publisher = {CCS},
   }

2024

1. ITEA

   From Text to Metadata: Automated Product Tagging with Python and Natural Language Processing

   Aayushi Verma and Omar Agha Khan

   Sep 2024

   Abs DOI Bib Video

   The Institute for Defense Analyses (IDA) produces a variety of research deliverables such as reports, memoranda, slides, and other formats for our sponsors. Summarizing keywords from these products quickly and for efficient retrieval of information on given research topics poses a challenge. IDA has numerous initiatives for tagging products with IDA-defined taxonomies of research terms, but this is a manual and time-consuming process and must be repeated periodically to tag newer products. To address this challenge, we developed a Python-based automated tagging pipeline. In this article, we introduce the mechanics of this pipeline, highlight current results, and discuss future applications for analyzing IDA’s research in terms of these tags.

   @article{Verma2024,
     title = {From Text to Metadata: Automated Product Tagging with Python and Natural Language Processing},
     author = {Verma, Aayushi and Khan, Omar Agha},
     year = {2024},
     month = sep,
     volume = {45},
     issue = {3},
     publisher = {The ITEA Journal of Test and Evaluation},
     doi = {10.61278/itea.45.3.1007},
     url = {https://itea.org/journals/volume-45-3/from-text-to-metadata-automated-product-tagging-with-python-and-natural-language-processing/},
   }

2023

1. ITEA

   I-TREE: A Tool for Characterizing Research Using Taxonomies

   Aayushi Verma

   Sep 2023

   Abs DOI Bib PDF Video

   The Institute for Defense Analyses (IDA) is developing a data strategy that implements data governance, data management, and data architecture practices and infrastructures. The data strategy leverages data to build trusted insights and establishes a data-centric culture. One component of the data strategy is a set of research taxonomies that describe and characterize research at IDA. We have created a dataset that consumes disparate data sources related to these taxonomies and unites them with metadata about research products and projects to create quantified analytics addressing questions about research at IDA. This dataset is curated and ingested by an interactive Shiny web application using R, which has been named I-TREE (IDA-Taxonomical Research Expertise Explorer). In this paper, I explain how we used data science to create I-TREE, which aids IDA in collecting new insights and making informed decisions.

   @article{Verma2023,
     title = {I-TREE: A Tool for Characterizing Research Using Taxonomies},
     author = {Verma, Aayushi},
     year = {2023},
     month = sep,
     volume = {44},
     issue = {3},
     publisher = {The ITEA Journal of Test and Evaluation},
     doi = {10.61278/itea.44.3.1001},
     url = {https://itea.org/journals/volume-44-3/I-TREE-A-Tool-for-characterizing-research-using-taxonomies/},
   }

2022

1. APJ

   Metallicity, ionization parameter, and pressure variations of H ii regions in the TYPHOON spiral galaxies: NGC 1566, NGC 2835, NGC 3521, NGC 5068, NGC 5236, and NGC 7793

   K. Grasha, Q. H. Chen, A. J. Battisti, and 12 more authors

   The Astrophysical Journal, Apr 2022

   Abs DOI Bib PDF

   We present a spatially resolved H ii region study of the gas-phase metallicity, ionization parameter, and interstellar medium (ISM) pressure maps of six local star-forming and face-on spiral galaxies from the TYPHOON program. Self-consistent metallicity, ionization parameter, and pressure maps are calculated simultaneously through an iterative process to provide useful measures of the local chemical abundance and its relation to localized ISM properties. We constrain the presence of azimuthal variations in metallicity by measuring the residual metallicity offset Δ(O/H) after subtracting the linear fits to the radial metallicity profiles. We, however, find weak evidence of azimuthal variations in most of the galaxies, with small (mean 0.03 dex) scatter. The galaxies instead reveal that H ii regions with enhanced and reduced abundances are found distributed throughout the disk. While the spiral pattern plays a role in organizing the ISM, it alone does not establish the relatively uniform azimuthal variations we observe. Differences in the metal abundances are more likely driven by the strong correlations with the local physical conditions. We find a strong and positive correlation between the ionization parameter and the local abundances as measured by the relative metallicity offset Δ(O/H), indicating a tight relationship between local physical conditions and their localized enrichment of the ISM. Additionally, we demonstrate the impact of unresolved observations on the measured ISM properties by rebinning the data cubes to simulate low-resolution (1 kpc) observations, typical of large IFU surveys. We find that the ionization parameter and ISM pressure diagnostics are impacted by the loss of resolution such that their measured values are larger relative to the measured values on sub-H ii region scales.

   @article{Grasha2022,
     doi = {10.3847/1538-4357/ac5ab2},
     url = {https://iopscience.iop.org/article/10.3847/1538-4357/ac5ab2},
     year = {2022},
     month = apr,
     publisher = {The American Astronomical Society},
     volume = {929},
     number = {2},
     pages = {118},
     author = {Grasha, K. and Chen, Q. H. and Battisti, A. J. and Acharyya, A. and Ridolfo, S. and Poehler, E. and Mably, S. and Verma, A. A. and Hayward, K. L. and Kharbanda, A. and Poetrodjojo, H. and Seibert, M. and Rich, J. A. and Madore, B. F. and Kewley, L. J.},
     title = {Metallicity, ionization parameter, and pressure variations of H ii regions in the TYPHOON spiral galaxies: NGC 1566, NGC 2835, NGC 3521, NGC 5068, NGC 5236, and NGC 7793},
     journal = {The Astrophysical Journal},
   }

2. PSJ

   The LCO Outbursting Objects Key Project: Overview and Year 1 Status

   Tim Lister, Michael S. P. Kelley, Carrie E. Holt, and 33 more authors

   The Planetary Science Journal, Jul 2022

   Abs DOI Bib PDF

   The LCO Outbursting Objects Key (LOOK) Project uses the telescopes of the Las Cumbres Observatory (LCO) Network to (1) systematically monitor a sample of previously discovered over the whole sky, to assess the evolutionary state of these distant remnants from the early solar system, and (2) use alerts from existing sky surveys to rapidly respond to and characterize detected outburst activity in all small bodies. The data gathered on outbursts helps to characterize each outburst’s evolution with time, helps to assess the frequency and magnitude distribution of outbursts in general, and contributes to the understanding of outburst processes and volatile distribution in the solar system. The LOOK Project exploits the synergy between current and future wide-field surveys such as ZTF, Pan-STARRS, and LSST, as well as rapid-response telescope networks such as LCO, and serves as an excellent test bed for what will be needed for the much larger number of objects coming from Rubin Observatory. We will describe the LOOK Project goals, the planning and target selection (including the use of NEOexchange as a Target and Observation Manager or “TOM”), and results from the first phase of observations, including the detection of activity and outbursts on the giant comet C/2014 UN271 (Bernardinelli–Bernstein) and the discovery and follow-up of 28 outbursts on 14 comets. Within these outburst discoveries, we present a high-cadence light curve of 7P/Pons–Winnecke with 10 outbursts observed over 90 days, a large outburst on 57P/duToit–Neujmin–Delporte, and evidence that comet P/2020 X1 (ATLAS) was in outburst when discovered.

   @article{Lister2022,
     doi = {10.3847/PSJ/ac7a31},
     url = {https://dx.doi.org/10.3847/PSJ/ac7a31},
     year = {2022},
     month = jul,
     publisher = {The American Astronomical Society},
     volume = {3},
     number = {7},
     pages = {173},
     author = {Lister, Tim and Kelley, Michael S. P. and Holt, Carrie E. and Hsieh, Henry H. and Bannister, Michele T. and Verma, Aayushi A. and Dobson, Matthew M. and Knight, Matthew M. and Moulane, Youssef and Schwamb, Megan E. and Bodewits, Dennis and Bauer, James and Chatelain, Joseph and Fernández-Valenzuela, Estela and Gardener, Daniel and Gyuk, Geza and Hammergren, Mark and Huynh, Ky and Jehin, Emmanuel and Kokotanekova, Rosita and Lilly, Eva and Hui, Man-To and McKay, Adam and Opitom, Cyrielle and Protopapa, Silvia and Ridden-Harper, Ryan and Schambeau, Charles and Snodgrass, Colin and Stoddard-Jones, Cai and Usher, Helen and Wierzchos, Kacper and Yanamandra-Fisher, Padma A. and Ye, Quanzhi and Project), (The LCO Outbursting Objects Key (LOOK) and Gomez, Edward and Greenstreet, Sarah},
     title = {The LCO Outbursting Objects Key Project: Overview and Year 1 Status},
     journal = {The Planetary Science Journal},
   }

2019

1. SS

   The Morphology of Galaxies

   Aayushi Verma

   In Southern Stars, Jun 2019

   Bib HTML PDF Video

   @inproceedings{Verma2019,
     author = {Verma, Aayushi},
     title = {The Morphology of Galaxies},
     booktitle = {Southern Stars},
     year = {2019},
     month = jun,
     volume = {58},
     pages = {7-10},
     url = {https://ui.adsabs.harvard.edu/abs/2019SouSt..58b...7V},
   }