arXiv subjects

Sarah Miller

Publications and source records attributed to Sarah Miller.

10 recordsLinked to original sources

arXiv2026

Ephemeris Refinement for Qatar-4 b, HAT-P-18 b, and CoRoT-1 b with Small Telescope and TESS Observations

We present updated transit timing measurements for three hot Jupiters (Qatar-4 b, HAT-P-18 b, and CoRoT-1 b) by leveraging data collected from the MicroObservatory Telescope Network, a network of small, robotic ground-based telescopes, and the NASA Transiting Exoplanet Survey Satellite (TESS). By combining these data with archival published results, we present the most precise orbital solutions to date for all three systems, allowing for precise transit time predictions for future missions. We report an updated mid-transit time for Qatar-4 b of 2458919.5838 $\pm$ 0.000089 $\mathrm{BJD}_{\mathrm{TDB}}$ and an updated orbital period of 1.80536560 $\pm$ 0.00000021 days. For HAT-P-18 b, we find a mid-transit time of 2459743.85340 $\pm$ 0.000022 $\mathrm{BJD}_{\mathrm{TDB}}$ and an updated orbital period of 5.50802957 $\pm$ 0.00000012 days. For CoRoT-1 b, we report a mid-transit time of 2456268.99083 $\pm$ 0.000099 $\mathrm{BJD}_{\mathrm{TDB}}$ and an updated orbital period of 1.50896846 $\pm$ 0.000000071 days. Our results demonstrate improvements over recently published ephemerides, with reductions of 36.4%, 4.35%, and 17.5% in mid-transit time uncertainties and 65.0%, 77.4%, and 16.9% in orbital period uncertainties for Qatar-4 b, HAT-P-18 b, and CoRoT-1 b, respectively. The results of this study improve the precision of future transit predictions and demonstrate the value of coordinated small-telescope monitoring (and citizen science initiatives) when updating the orbital parameters of hot Jupiters.

astro-ph.EP↗

arXiv2026

GazeXPErT: An Expert Eye-tracking Dataset for Interpretable and Explainable AI in Oncologic FDG-PET/CT Scans

[18F]FDG-PET/CT is a cornerstone imaging modality for guiding oncology therapies, yet human expert shortages necessitate more efficient diagnostic aids. While standalone AI models for automatic lesion detection exist, clinical translation remains hindered by AI explainability, reliability, and workflow integration. Meanwhile, human-computer-interaction in radiology remain limited to keyboard, mouse and voice, ignoring experts' faster, natural gaze signal. We present GazeXPErT, a 4D eye-tracking dataset with annotated expert decision windows for tumor detection and measurement on 346 dual-read FDG-PET/CTs. The dataset contributes 9,030 gaze-to-lesion trajectories derived from 3,948 minutes of 60 Hz eye-tracking data, rendered in COCO-style format. GazeXPErT captures experts' visual reasoning patterns when adjudicating suspicious lesions. It aims to facilitate development of trusted, explainable and interactive AI models through understanding expert gaze patterns. Baseline feasibility experiments suggest salient signal is extractable from routinely collected expert gaze (3D nnU-Net Dice: 0.6819 versus 0.6008 without), that gaze-trained vision transformers may aid dynamic lesion localization (74.95% predicted gaze closer to tumor), and that experts' intent may be predictable from raw gaze (Accuracy 67.53%, AUROC 0.747).

eess.IV↗

arXiv2025

Sub-diffraction-resolved spatial distribution of emitting excitons in STM-induced luminescence of 2D semiconductors via Richardson-Lucy deconvolution

Using scanning tunneling microscopy-induced luminescence (STML), the optical properties of two-dimensional (2D) semiconductors may be investigated at the nanoscale. This is possible because the tunneling current under the tip is an extremely localized electrical excitation source. However, in most STML applications, the spatial distribution of the emission relative to the excitation point is unresolved. Yet this distribution contains key information about how the interaction of excitons with injected charge carriers affects the luminescence of these materials, and about exciton transport. Resolving this spatial distribution at the nanoscale is relevant both for a fundamental understanding of exciton physics and for device applications; yet it remains a significant challenge. In this work, we resolve the spatial distribution of the emission beyond the diffraction limit of light by deconvolving real-space optical microscopy images of the STML using an iterative algorithm, i.e., Richardson-Lucy (RL) deconvolution. To showcase this technique, we apply it to the STML of monolayer tungsten diselenide ($\mathrm{WSe_2}$) and tungsten disulfide ($\mathrm{WS_2}$). Thus, we highlight hitherto ignored or misunderstood aspects of STML on 2D semiconductors related to exciton and charge carrier transport, namely the dependence of the spatial distribution of emission on the tunnel current setpoint and the origin of the emission from hot spots located micrometers from the excitation source.

cond-mat.mes-hall↗

arXiv2025

On sums of $\mathscr{P}$-free forms under mis\`ere play

Milley and Renault proved an interesting characterisation of invertible elements in the dead-ending universe: they are the games with no subpositions of outcome $\mathscr{P}$ (the '$\mathscr{P}$-free' games). We generalise their approach to obtain a stronger result and show in particular that the set of $\mathscr{P}$-free blocking games is closed under addition, which yields that every $\mathscr{P}$-free blocking game is invertible modulo the blocking universe. This has consequences for the invertible subgroups of various other mis\`ere monoids.

math.CO↗

arXiv2023

ASL Champ!: A Virtual Reality Game with Deep-Learning Driven Sign Recognition

We developed an American Sign Language (ASL) learning platform in a Virtual Reality (VR) environment to facilitate immersive interaction and real-time feedback for ASL learners. We describe the first game to use an interactive teaching style in which users learn from a fluent signing avatar and the first implementation of ASL sign recognition using deep learning within the VR environment. Advanced motion-capture technology powers an expressive ASL teaching avatar within an immersive three-dimensional environment. The teacher demonstrates an ASL sign for an object, prompting the user to copy the sign. Upon the user's signing, a third-party plugin executes the sign recognition process alongside a deep learning model. Depending on the accuracy of a user's sign production, the avatar repeats the sign or introduces a new one. We gathered a 3D VR ASL dataset from fifteen diverse participants to power the sign recognition model. The proposed deep learning model's training, validation, and test accuracy are 90.12%, 89.37%, and 86.66%, respectively. The functional prototype can teach sign language vocabulary and be successfully adapted as an interactive ASL learning platform in VR.

cs.HC↗

arXiv2021

Blending Knowledge in Deep Recurrent Networks for Adverse Event Prediction at Hospital Discharge

Deep learning architectures have an extremely high-capacity for modeling complex data in a wide variety of domains. However, these architectures have been limited in their ability to support complex prediction problems using insurance claims data, such as readmission at 30 days, mainly due to data sparsity issue. Consequently, classical machine learning methods, especially those that embed domain knowledge in handcrafted features, are often on par with, and sometimes outperform, deep learning approaches. In this paper, we illustrate how the potential of deep learning can be achieved by blending domain knowledge within deep learning architectures to predict adverse events at hospital discharge, including readmissions. More specifically, we introduce a learning architecture that fuses a representation of patient data computed by a self-attention based recurrent neural network, with clinically relevant features. We conduct extensive experiments on a large claims dataset and show that the blended method outperforms the standard machine learning approaches.

cs.LG↗

arXiv2021

Question-Driven Design Process for Explainable AI User Experiences

A pervasive design issue of AI systems is their explainability--how to provide appropriate information to help users understand the AI. The technical field of explainable AI (XAI) has produced a rich toolbox of techniques. Designers are now tasked with the challenges of how to select the most suitable XAI techniques and translate them into UX solutions. Informed by our previous work studying design challenges around XAI UX, this work proposes a design process to tackle these challenges. We review our and related prior work to identify requirements that the process should fulfill, and accordingly, propose a Question-Driven Design Process that grounds the user needs, choices of XAI techniques, design, and evaluation of XAI UX all in the user questions. We provide a mapping guide between prototypical user questions and exemplars of XAI techniques to reframe the technical space of XAI, also serving as boundary objects to support collaboration between designers and AI engineers. We demonstrate it with a use case of designing XAI for healthcare adverse events prediction, and discuss lessons learned for tackling design challenges of AI systems.

cs.HC↗

arXiv2020

A Canonical Architecture For Predictive Analytics on Longitudinal Patient Records

Many institutions within the healthcare ecosystem are making significant investments in AI technologies to optimize their business operations at lower cost with improved patient outcomes. Despite the hype with AI, the full realization of this potential is seriously hindered by several systemic problems, including data privacy, security, bias, fairness, and explainability. In this paper, we propose a novel canonical architecture for the development of AI models in healthcare that addresses these challenges. This system enables the creation and management of AI predictive models throughout all the phases of their life cycle, including data ingestion, model building, and model promotion in production environments. This paper describes this architecture in detail, along with a qualitative evaluation of our experience of using it on real world problems.

cs.LG↗

arXiv2020

Questioning the AI: Informing Design Practices for Explainable AI User Experiences

A surge of interest in explainable AI (XAI) has led to a vast collection of algorithmic work on the topic. While many recognize the necessity to incorporate explainability features in AI systems, how to address real-world user needs for understanding AI remains an open question. By interviewing 20 UX and design practitioners working on various AI products, we seek to identify gaps between the current XAI algorithmic work and practices to create explainable AI products. To do so, we develop an algorithm-informed XAI question bank in which user needs for explainability are represented as prototypical questions users might ask about the AI, and use it as a study probe. Our work contributes insights into the design space of XAI, informs efforts to support design practices in this space, and identifies opportunities for future XAI work. We also provide an extended XAI question bank and discuss how it can be used for creating user-centered XAI.

cs.HC↗

arXiv2015

Nebular and Stellar Dust Extinction Across the Disk of Emission-Line Galaxies on Small (KPC) Scales

We investigate resolved kpc-scale stellar and nebular dust distribution in eight star-forming galaxies at z~0.4 in the GOODS fields. Constructing the observed Spectral Energy Distributions (SEDs) per pixel, based on seven bands photometric data from HST/ACS and WFC3, we performed pixel-by-pixel SED fits to population synthesis models and estimated small-scale distribution of stellar dust extinction. We use Halpha / Hbeta nebular emission line ratios from Keck/DEIMOS high resolution spectra at each spatial resolution element to measure the amount of attenuation faced by ionized gas at different radii from the center of galaxies. We find a good agreement between the integrated and median of resolved color excess measurements in our galaxies. The ratio of integrated nebular to stellar dust extinction is always greater than unity, but does not show any trend with stellar mass or star formation rate. We find that inclination plays an important role in the variation of the nebular to stellar excess ratio. The stellar color excess profiles are found to have higher values at the center compared to outer parts of the disk. However, for lower mass galaxies, a similar trend is not found for the nebular color excess. We find that the nebular color excess increases with stellar mass surface density. This explains the absence of radial trend in the nebular color excess in lower mass galaxies which lack a large radial variation of stellar mass surface density. Using standard conversions of star formation rate surface density to gas mass surface density, and the relation between dust mass surface density and color excess, we find no significant variation in the dust to gas ratio in regions with high gas mass surface densities, over the scales probed in this study.

astro-ph.GA↗