arXiv subjects

Sam Vaughan

Publications and source records attributed to Sam Vaughan.

7 recordsLinked to original sources

arXiv2026

Media Integrity and Authentication: Status, Directions, and Futures

We provide background on emerging challenges and future directions with media integrity and authentication methods, focusing on distinguishing AI-generated media from authentic content captured by cameras and microphones. We evaluate several approaches, including provenance, watermarking, and fingerprinting. After defining each method, we analyze three representative technologies: cryptographically secured provenance, imperceptible watermarking, and soft-hash fingerprinting. We analyze how these tools operate across modalities and evaluate relevant threat models, attack categories, and real-world workflows spanning capture, editing, distribution, and verification. We consider sociotechnical reversal attacks that can invert integrity signals, making authentic content appear synthetic and vice versa, highlighting the value of verification systems that are resilient to both technical and psychosocial manipulation. Finally, we outline techniques for delivering high-confidence provenance authentication, including directions for strengthening edge-device security using secure enclaves.

cs.CR↗

arXiv2025

The MAGPI Survey: the kinematic morphology-density relation (or lack thereof) and the Hubble sequence at $z\sim0.3$

This work presents visual morphological and dynamical classifications for 637 spatially resolved galaxies, most of which are at intermediate redshift ($z\sim0.3$), in the Middle-Ages Galaxy Properties with Integral field spectroscopy (MAGPI) Survey. For each galaxy, we obtain a minimum of 11 independent visual classifications by knowledgeable classifiers. We use an extension of the standard Dawid-Skene bayesian model introducing classifier-specific confidence parameters and galaxy-specific difficulty parameters to quantify classifier confidence and infer reliable statistical confidence estimates. Selecting sub-samples of 86 bright ($r<20$ mag) high-confidence ($>0.98$) morphological classifications at redshifts ($0.2 \le z \le0.4$), we confirm the full range of morphological types is represented in MAGPI as intended in the survey design. Similarly, with a sub-sample of 82 bright high-confidence stellar kinematic classifications, we find that the rotating and non-rotating galaxies seen at low redshift are already in place at intermediate redshifts. We \textit{do not} find evidence that the kinematic morphology-density relation seen at $z\sim0$ is established at $z\sim0.3$. We suggest that galaxies without obvious stellar rotation are dynamically pre-processed sometime before $z\sim0.3$ within lower mass groups before joining denser environments.

astro-ph.GA↗

arXiv2025

Lessons From Red Teaming 100 Generative AI Products

In recent years, AI red teaming has emerged as a practice for probing the safety and security of generative AI systems. Due to the nascency of the field, there are many open questions about how red teaming operations should be conducted. Based on our experience red teaming over 100 generative AI products at Microsoft, we present our internal threat model ontology and eight main lessons we have learned: 1. Understand what the system can do and where it is applied 2. You don't have to compute gradients to break an AI system 3. AI red teaming is not safety benchmarking 4. Automation can help cover more of the risk landscape 5. The human element of AI red teaming is crucial 6. Responsible AI harms are pervasive but difficult to measure 7. LLMs amplify existing security risks and introduce new ones 8. The work of securing AI systems will never be complete By sharing these insights alongside case studies from our operations, we offer practical recommendations aimed at aligning red teaming efforts with real world risks. We also highlight aspects of AI red teaming that we believe are often misunderstood and discuss open questions for the field to consider.

cs.AI↗

arXiv2023

Validating full-spectrum fitting with a synthetic integral-field spectroscopic observation of the Milky Way

Ongoing deep IFS observations of disk galaxies provide opportunities for comparison with the Milky Way (MW) to understand galaxy evolution. However, such comparisons are marred by many challenges such as selection effects, differences in observations and methodology, and proper validation of full-spectrum fitting methods. In this study, we present a novel code GalCraft to address these challenges by generating mock IFS data cubes of the MW using simple stellar population models and a mock MW stellar catalog derived from E-Galaxia. We use the widely adopted full-spectrum fitting code pPXF to investigate the ability to recover kinematics and stellar populations for an edge-on mock MW IFS observation. We confirm that differences in kinematics, mean age, [M/H], and [$\alpha$/Fe] between thin and thick disks can be distinguished. However, the age distribution is overestimated in the ranges between 2 - 4 and 12 - 14 Gyr compared to the expected values. This is likely due to the age spacing and degeneracy of SSP templates. We find systematic offsets in the recovered kinematics due to insufficient spectral resolution and the variation of line-of-sight velocity distribution with age and [M/H]. With future higher resolution and multi-[$\alpha$/Fe] SSP templates, GalCraft will be useful to validate key signatures such as [$\alpha$/Fe]-[M/H] distribution at different $R$ and $|z|$ and potentially infer radial migration and kinematic heating efficiency to study detailed chemodynamical evolution of MW-like galaxies.

astro-ph.GA↗

arXiv2023

The MAGPI Survey: Impact of environment on the total internal mass distribution of galaxies in the last 5 Gyr

We investigate the impact of environment on the internal mass distribution of galaxies using the Middle Ages Galaxy Properties with Integral field spectroscopy (MAGPI) survey. We use 2D resolved stellar kinematics to construct Jeans dynamical models for galaxies at mean redshift $z \sim 0.3$, corresponding to a lookback time of $3-4$ Gyr. The internal mass distribution for each galaxy is parameterised by the combined mass density slope $\gamma$ (baryons $+$ dark matter), which is the logarithmic change of density with radius. We use a MAGPI sample of 28 galaxies from low-to-mid density environments and compare to density slopes derived from galaxies in the high density Frontier Fields clusters in the redshift range $0.29 <z < 0.55$, corresponding to a lookback time of $\sim 5$ Gyr. We find a median density slope of $\gamma = -2.22 \pm 0.05$ for the MAGPI sample, which is significantly steeper than the Frontier Fields median slope ($\gamma = -2.01 \pm 0.04$), implying the cluster galaxies are less centrally concentrated in their mass distribution than MAGPI galaxies. We also compare to the distribution of density slopes from galaxies in Atlas3D at $z \sim 0$, because the sample probes a similar environmental range as MAGPI. The Atlas3D median total slope is $\gamma = -2.25 \pm 0.02$, consistent with the MAGPI median. Our results indicate environment plays a role in the internal mass distribution of galaxies, with no evolution of the slope in the last 3-4 Gyr. These results are in agreement with the predictions of cosmological simulations.

astro-ph.GA↗

arXiv2023

The SAMI Survey: Evidence for dynamical coupling of ionised gas and young stellar populations

We explore local and global dynamical differences between the kinematics of ionised gas and stars in a sample of galaxies from Data Release 3 of the SAMI Galaxy Survey. We find better agreement between local (i.e., comparing on a spaxel-to-spaxel basis) velocities and dispersion of gas and stars in younger systems as with previous work on the asymmetric drift in galaxies, suggesting that the dynamics of stars and ionised gas are initially coupled. The intrinsic scatter around the velocity and dispersion relations increases with increasing stellar age and mass, suggesting that subsequent mechanisms such as internal processes, divergent star formation and assembly histories also play a role in setting and altering the dynamics of galaxies. The global (flux-weighted) dynamical support of older galaxies is hotter than in younger systems. We find that the ionised gas in galaxies is almost always dynamically colder than the stars with a steeper velocity gradient. In absolute terms, the local difference in velocity dispersion is more pronounced than the local difference in velocity, possibly reflecting inherent differences in the impact of turbulence, inflow and/or feedback on gas compared to stars. We suggest how these findings may be taken into account when comparing high and low redshift galaxy samples to infer dynamical evolution.

astro-ph.GA↗

arXiv2018

SDSS-IV MaNGA: The Spatially Resolved Stellar Initial Mass Function in $\sim$400 Early-Type Galaxies

MaNGA provides the opportunity to make precise spatially resolved measurements of the IMF slope in galaxies owing to its unique combination of spatial resolution, wavelength coverage and sample size. We derive radial gradients in age, element abundances and IMF slope analysing optical and near-infrared absorption features from stacked spectra out to the half-light radius of 366 early-type galaxies with masses $9.9 - 10.8\;\log M/M_{\odot}$. We find flat gradients in age and [$\alpha$/Fe] ratio, as well as negative gradients in metallicity, consistent with the literature. We further derive significant negative gradients in the [Na/Fe] ratio with galaxy centres being well enhanced in Na abundance by up to 0.5 dex. Finally, we find a gradient in IMF slope with a bottom-heavy IMF in the centre (typical mass excess factor of 1.5) and a Milky Way-type IMF at the half-light radius. This pattern is mass-dependent with the lowest mass galaxies in our sample featuring only a shallow gradient around a Milky Way IMF. Our results imply the local IMF-$\sigma$ relation within galaxies to be even steeper than the global relation and hint towards the local metallicity being the dominating factor behind the IMF variations. We also employ different stellar population models in our analysis and show that a radial IMF gradient is found independently of the stellar population model used. A similar analysis of the Wing-Ford band provides inconsistent results and further evidence of the difficulty in measuring and modelling this particular feature.

astro-ph.GA↗