arXiv subjects

Benjamin Thompson

Publications and source records attributed to Benjamin Thompson.

12 recordsLinked to original sources

arXiv2026

Measuring the Prevalence of Policy Violating Content with ML Assisted Sampling and LLM Labeling

Content safety teams need metrics that reflect what users actually experience, not only what is reported. We study prevalence: the fraction of user views (impressions) that went to content violating a given policy on a given day. Accurate prevalence measurement is challenging because violations are often rare and human labeling is costly, making frequent, platform-representative studies slow. We present a design-based measurement system that (i) draws daily probability samples from the impression stream using ML-assisted weights to concentrate label budget on high-exposure and high-risk content while preserving unbiasedness, (ii) labels sampled items with a multimodal LLM governed by policy prompts and gold-set validation, and (iii) produces design-consistent prevalence estimates with confidence intervals and dashboard drilldowns. A key design goal is one global sample with many pivots: the same daily sample supports prevalence by surface, viewer geography, content age, and other segments through post-stratified estimation. We describe the statistical estimators, variance and confidence interval construction, label-quality monitoring, and an engineering workflow that makes the system configurable across policies.

cs.LG↗

arXiv2026

Test-retest Reliability of Psychophysical Tasks Using Structured Light-Induced Entoptic Phenomena

Structured light (SL)-induced polarization perception presents a promising method for the early detection of macular diseases such as age-related macular degeneration. We investigated the test-retest reliability of a psychophysical task using SL-based stimuli to induce entoptic patterns in individuals with healthy vision. Twenty-eight participants underwent thorough eye examinations to confirm they had healthy eyes and good vision (logMAR BCVA ~0.00). Of these, 25 participants (n=50 eyes) aged 21 to 75 completed two identical tasks separated by 1 to 14 days. Using SL-based stimuli that produced a rotating entoptic pattern containing 22 azimuthal brushes, we measured the retinal eccentricity threshold $R_T$ at which participants could reliably identify the direction of rotation by varying the size of a central obstruction. This threshold reflects the visual angle of the pattern for each participant. We calculated the reliability coefficient (intraclass correlation coefficient, ICC) using a two-way mixed-effects model and conducted a Bland-Altman analysis to assess test-retest reliability. The ICC was 0.83 [95% CI: 0.62 - 0.93] for right eyes (RE) and 0.93 [95% CI: 0.84 - 0.97] for left eyes (LE), indicating good reliability. The Bland-Altman analysis showed a mean difference of -0.32{\deg} (SD: 1.51) and -0.12{\deg} (SD: 1.03) for RE and LE respectively between the first and second sessions, with limits of agreement ranging from -3.28{\deg} to 2.64{\deg} and -2.14{\deg} to 1.89{\deg} for RE and LE respectively, confirming strong agreement and no significant bias. These results demonstrate that SL-based psychophysical tasks are a reliable method for assessing polarization perception, potentially improving screening for diseases affecting macular health.

physics.med-ph↗

arXiv2025

Topological Expansion of Boehm's Brushes via Structured Light

We report a novel entoptic phenomenon in which the classical two-lobed Boehm's brushes are transformed into a multi-lobed structure by projecting spin-orbit coupled light onto the human retina. These structured beams, composed of non-separable superpositions of circular polarization and orbital angular momentum (OAM), produce azimuthally modulated entoptic patterns through polarization-dependent scattering in the retina. Unlike Haidinger's brushes, which arise from dichroic absorption in the macula, the observed effect is driven by angular variations in scattering strength relative to the local polarization direction. In regions where scattering centers exhibit polarization orientations that converge toward a common point, their combined contributions reinforce one another, producing brighter and more sharply defined entoptic lobes whose number and orientation vary systematically with the topology of the spin-orbit stimulus. Psychophysical measurements across retinal eccentricities from 0.5$^\circ$ to 4$^\circ$ in eleven participants revealed that contrast detection thresholds decreased exponentially with eccentricity, consistent with polarization-sensitive scattering by isotropic structures in the non-foveal retinal regions. From the psychophysical fits, the mean eccentricity at which the entoptic pattern reached a 50 % threshold was $r_{50} = 1.03^\circ$ with a 95 % confidence interval of [0.72, 1.34]$^\circ$, indicating that the spin-orbit-induced entoptic structure becomes perceptually robust at approximately 1$^\circ$ retinal eccentricity. Together, these findings demonstrate that spin-orbit light modulates scattering-based visual phenomena in previously unrecognized ways, enabling new approaches for probing retinal structure and visual processing using topological features of light.

physics.optics↗

arXiv2024

Characterizing the circularly-oriented macular pigment using spatiotemporal sensitivity to structured light entoptic phenomena

The macular pigment (MP) in the radially-oriented Henle fibers that surround the foveola enables the ability to perceive the orientation of polarized blue light through an entoptic phenomena known as the Haidinger's brush. The MP has been linked to eye diseases and central field dysfunctions, most notably age-related macular degeneration (AMD), a globally leading cause of irreversible blindness. Recent integration of structured light techniques into vision science has allowed for the development of more selective and versatile entoptic probes of eye health that provide interpretable thresholds. For example, it enabled the use of variable spatial frequencies and arbitrary obstructions in the presented stimuli. Additionally, it expanded the 2{\deg} retinal eccentricity extent of the Haidinger's brush to 5{\deg} for a similar class of fringe-based stimuli. In this work, we develop a spatiotemporal sensitivity model that maps perceptual thresholds of entoptic phenomenon to the underlying MP structure that supports its perception. We therefore selectively characterize the circularly-oriented macular pigment optical density (coMPOD) rather than total MPOD as typically measured, providing an additional quantification of macular health. A study was performed where the retinal eccentricity thresholds were measured for five structured light stimuli with unique spatiotemporal frequencies. The results from fifteen healthy young participants indicate that the coMPOD is inversely proportional to retinal eccentricity in the range of 1.5{\deg} to 5.5{\deg}. Good agreement between the model and the collected data is found with a Pearson $\chi^2$ fit statistic of 0.06. The presented techniques can be applied in novel early diagnostic tests for a variety of diseases related to macular degeneration such as AMD, macular telangiectasia, and pathological myopia.

physics.med-ph↗

arXiv2021

High precision control and deep learning-based corn stand counting algorithms for agricultural robot

This paper presents high precision control and deep learning-based corn stand counting algorithms for a low-cost, ultra-compact 3D printed and autonomous field robot for agricultural operations. Currently, plant traits, such as emergence rate, biomass, vigor, and stand counting, are measured manually. This is highly labor-intensive and prone to errors. The robot, termed TerraSentia, is designed to automate the measurement of plant traits for efficient phenotyping as an alternative to manual measurements. In this paper, we formulate a Nonlinear Moving Horizon Estimator (NMHE) that identifies key terrain parameters using onboard robot sensors and a learning-based Nonlinear Model Predictive Control (NMPC) that ensures high precision path tracking in the presence of unknown wheel-terrain interaction. Moreover, we develop a machine vision algorithm designed to enable an ultra-compact ground robot to count corn stands by driving through the fields autonomously. The algorithm leverages a deep network to detect corn plants in images, and a visual tracking model to re-identify detected objects at different time steps. We collected data from 53 corn plots in various fields for corn plants around 14 days after emergence (stage V3 - V4). The robot predictions have agreed well with the ground truth with $C_{robot}=1.02 \times C_{human}-0.86$ and a correlation coefficient $R=0.96$. The mean relative error given by the algorithm is $-3.78\%$, and the standard deviation is $6.76\%$. These results indicate a first and significant step towards autonomous robot-based real-time phenotyping using low-cost, ultra-compact ground robots for corn and potentially other crops.

cs.RO↗

arXiv2021

The Binary INformation from Open Clusters Using SEDs (BINOCS) Project: Reliable Photometric Mass Determinations of Binary Star Systems in Clusters

We introduce a new binary detection technique, Binary INformation from Open Clusters using SEDs (binocs), which we show is able to determine reliable stellar multiplicity and masses over a much larger mass range than current approaches. This new technique determines accurate component masses of binary and single systems of the open clusters main sequence by comparing observed magnitudes from multiple photometric filters to synthetic star spectral energy distributions (SEDs) allowing systematically probing the binary population for low mass stars in clusters for 8 well-studied open clusters. We provide new deep, infrared photometric catalogs (1.2 - 8.0 microns) for the key open clusters NGC 1960 (M36), NGC 2099 (M37), NGC 2420, and NGC2682 (M67), using observation from NOAO/NEWFIRM and Spitzer}/IRAC. Using these deep multi-wavelength catalogs, the binocs method is applied to these clusters to determine accurate component masses for unresolved cluster binaries. We explore binary fractions as a function of cluster age, Galactic location and metallicity.

astro-ph.SR↗

arXiv2018

The Open Cluster Chemical Abundances and Mapping Survey: II. Precision Cluster Abundances for APOGEE using SDSS DR14

The Open Cluster Chemical Abundances and Mapping (OCCAM) survey aims to produce a comprehensive, uniform, infrared-based spectroscopic dataset for hundreds of open clusters, and to constrain key Galactic dynamical and chemical parameters from this sample. This second contribution from the OCCAM survey presents analysis of 259 member stars with [Fe/H] determinations in 19 open clusters, using Sloan Digital Sky Survey Data Release 14 (SDSS/DR14) data from the Apache Point Observatory Galactic Evolution Experiment (APOGEE) and ESA Gaia. This analysis, which includes clusters with R_{GC} ranging from 7 to 13 kpc, measures an [Fe/H] gradient of -0.061 \pm 0.004 dex/kpc. We also confirm evidence of a significant positive gradient in the \alpha-elements ([O/Fe], [Mg/Fe], and [Si/Fe]) and present evidence for a significant negative gradient in iron-peak elements ([Mn/Fe] and [Ni/Fe]).

astro-ph.GA↗

arXiv2017

Khovanov complexes of rational tangles

We show that the Khovanov complex of a rational tangle has a very simple representative whose backbone of non-zero morphisms forms a zig-zag. Furthermore, this minimal complex can be computed quickly by an inductive algorithm. (For example, we calculate $Kh(8_2)$ by hand.) We find that the bigradings of the subobjects in these minimal complexes can be described by matrix actions, which after a change of basis is the reduced Burau representation of $B_3$.

math.GT↗

arXiv2016

Chemical abundance gradients from open clusters in the Milky Way disk: results from the APOGEE survey

Metallicity gradients provide strong constraints for understanding the chemical evolution of the Galaxy. We report on radial abundance gradients of Fe, Ni, Ca, Si, and Mg obtained from a sample of 304 red-giant members of 29 disk open clusters, mostly concentrated at galactocentric distances between ~8 - 15 kpc, but including two open clusters in the outer disk. The observations are from the APOGEE survey. The chemical abundances were derived automatically by the ASPCAP pipeline and these are part of the SDSS III Data Release 12. The gradients, obtained from least squares fits to the data, are relatively flat, with slopes ranging from -0.026 to -0.033 dex/kpc for the alpha-elements [O/H], [Ca/H], [Si/H] and [Mg/H] and -0.035 dex/kpc and -0.040 dex/kpc for [Fe/H] and [Ni/H], respectively. Our results are not at odds with the possibility that metallicity ([Fe/H]) gradients are steeper in the inner disk (R_GC ~7 - 12 kpc) and flatter towards the outer disk. The open cluster sample studied spans a significant range in age. When breaking the sample into age bins, there is some indication that the younger open cluster population in our sample (log age < 8.7) has a flatter metallicity gradient when compared with the gradients obtained from older open clusters.

astro-ph.GA↗

arXiv2015

Binary Information from Open Clusters Using SEDS (BINOCS) Project: The Dynamical Evolution of the Binary Populations in Cluster Environments

Studying the internal dynamics of stellar clusters is conducted primarily through N-Body simulations. One of the major inputs into N-Body simulations is the binary star frequency and mass distribution, which is currently constrained by relations derived from field binary stars. However to truly understand how clustered environments evolve, binary data from within star clusters is needed including masses. Detailed information on binaries masses, primary and secondary, in star clusters has been limited to date. The primary technique currently available has been radial velocity surveys that are limited in depth. Using previous two-band photometry-based studies that may cover different mass ranges produce potentially discrepant interpretations of the observed binary population. We introduce a new binary detection method, Binary INformation from Open Clusters Using SEDs (BINOCS) that covers the wide mass range needed to improve cluster N-body simulation inputs and comparisons. Using newly-observed multi-wavelength photometric catalogs (0.3 - 8 microns) of the key open clusters with a range of ages, we can show that the BINOCS method determines accurate binary component masses for unresolved cluster binaries through comparison to available RV-based studies. Using this method, we present results on the dynamical evolution of binaries from 0.4 - 2.5 solar masses within five prototypical clusters, spaning 30 Myr to 3.5 Gyr, and how the binary populations evolve as a function of mass.

astro-ph.SR↗

arXiv2015

The Apache Point Observatory Galactic Evolution Experiment (APOGEE)

The Apache Point Observatory Galactic Evolution Experiment (APOGEE), one of the programs in the Sloan Digital Sky Survey III (SDSS-III), has now completed its systematic, homogeneous spectroscopic survey sampling all major populations of the Milky Way. After a three year observing campaign on the Sloan 2.5-m Telescope, APOGEE has collected a half million high resolution (R~22,500), high S/N (>100), infrared (1.51-1.70 microns) spectra for 146,000 stars, with time series information via repeat visits to most of these stars. This paper describes the motivations for the survey and its overall design---hardware, field placement, target selection, operations---and gives an overview of these aspects as well as the data reduction, analysis and products. An index is also given to the complement of technical papers that describe various critical survey components in detail. Finally, we discuss the achieved survey performance and illustrate the variety of potential uses of the data products by way of a number of science demonstrations, which span from time series analysis of stellar spectral variations and radial velocity variations from stellar companions, to spatial maps of kinematics, metallicity and abundance patterns across the Galaxy and as a function of age, to new views of the interstellar medium, the chemistry of star clusters, and the discovery of rare stellar species. As part of SDSS-III Data Release 12, all of the APOGEE data products are now publicly available.

astro-ph.IM↗

arXiv2013

The Open Cluster Chemical Analysis and Mapping Survey: Local Galactic Metallicity Gradient with APOGEE using SDSS DR10

The Open Cluster Chemical Analysis and Mapping (OCCAM) Survey aims to produce a comprehensive, uniform, infrared-based dataset for hundreds of open clusters, and constrain key Galactic dynamical and chemical parameters from this sample. This first contribution from the OCCAM survey presents analysis of 141 members stars in 28 open clusters with high-resolution metallicities derived from a large uniform sample collected as part of the SDSS-III/Apache Point Observatory Galactic Evolution Experiment (APOGEE). This sample includes the first high-resolution metallicity measurements for 22 open clusters. With this largest ever uniformly observed sample of open cluster stars we investigate the Galactic disk gradients of both [M/H] and [alpha/M]. We find basically no gradient across this range in [alpha/M], but [M/H] does show a gradient for R_{GC} < 10 kpc and a significant flattening beyond R_{GC} = 10 kpc. In particular, whereas fitting a single linear trend yields an [M/H] gradient of -0.09 +/- 0.03$ dex/kpc --- similar to previously measure gradients inside 13 kpc --- by independently fitting inside and outside 10 kpc separately we find a significantly steeper gradient near the Sun (7.9 <= R_{GC} <= 10) than previously found (-0.20 +/- 0.08 dex/kpc) and a nearly flat trend beyond 10 kpc (-0.02 +/- 0.09 dex/kpc).

astro-ph.GA↗