arXiv subjects

Minjung Park

Publications and source records attributed to Minjung Park.

At least 19 recordsLinked to original sources

arXiv2026

A Dusty Quenching-candidate AGN host at z=5.7: massive quiescent galaxies may quench already during the dust-obscured phase

We present a spectro-photometric analysis of RUBIES-EGS-9809, a broad-line AGN host at z=5.7 with a Balmer break and a dominant point source. The source is dust reddened ($A_V>$3 mag), shows hot- and cold-dust emission, and is X-ray Compton thick, with X-ray-to-bolometric ratio consistent with luminous AGN. The large $A_V$ agrees with strong NaI absorption (EW=17$\pm 2 \AA$), possibly from a neutral-gas outflow (3 $\sigma$), while broad [O III] traces an ionized outflow, suggesting AGN feedback may be affecting the host. We estimate a stellar mass log(M*/M$_\odot$)=10.7$\pm$0.3, Lbol$\sim4\times10^{46}$ erg s$^{-1}$ , and black-hole mass log(M$_{BH}/M_\odot)\sim8.1$ (subject to large systematics). The resolved Balmer break (0.1 arcsec, 0.6 kpc) is consistent with a stellar origin, implying the galaxy is already old in stars while still heavily obscured. This shows that massive galaxies can host evolved stellar populations during their compact, dust-obscured phase, consistent with formation in an earlier dusty starburst. SED modelling suggests a declining SFR, but low-resolution spectroscopy alone cannot constrain the recent SFR, so we treat the source as a strong quenching candidate rather than a secure post-starburst system. Because quenching follows the decline in SFR, star-formation-driven outflows are disfavoured, leaving AGN feedback as a plausible driver. We speculate that such rapid feedback ($\sim$100 Myr timescale) may precipitate quenching while the post-quenching phase stays dust-enshrouded, concealing the UV-bright quenched phase that would otherwise be easily detected, if dust free. By the time dust clears, the UV-luminous stars may have already dwindled, leaving a classic dust-free quiescent galaxy. This 'dusty path' to quiescence may explain the lack of intermediate-age quenched progenitors (ages 50-100 Myr) linking dusty starbursts to massive quiescent systems at z=3-7.

astro-ph.GA↗

arXiv2026

When Are LLM Inferences Acceptable? User Reactions and Control Preferences for Inferred Personal Information

Ask ChatGPT about vacation planning, and it may infer your income. Ask it about medication, and it may infer your medical history. Because such inferences can expose more information than users intend to reveal, prior work argues that they are a defining privacy risk of LLM-based systems. Yet prior work has mostly shown that LLMs can make potentially violating inferences, not how users experience those inferences nor what controls users may want governing their use. We built the Reflective Layer, a visualization tool that surfaces example unstated inferences from users' own ChatGPT histories, and used it in a mixed-methods study with 18 regular ChatGPT users evaluating 215 surfaced inferences from their own conversations. Counterintuitively, participants reacted more strongly with curiosity and interest rather than distress and concern. Discomfort arose mainly when inferences felt misrepresentative of the user or misaligned with expected use. Participants were also markedly less comfortable with advertisers and third-party applications using those inferences than with platform providers. These findings suggest that the acceptability of LLM inferences is governed not only by its content, but by context-sensitive norms around how they are generated, retained within the platform, and transmitted beyond it.

cs.HC↗

arXiv2026

Probing neutral outflows in z ~ 2 galaxies using JWST observations of Ca II H and K absorption lines

Using deep JWST/NIRSpec spectra from the Blue Jay survey, we perform the first systematic investigation of neutral gas content in massive galaxies at Cosmic Noon based on the Ca II H, K absorption lines. We analyze a sample of 9 galaxies at 1.8 < z < 2.8 with stellar masses > 10.6, for which we detect neutral gas absorption both in Ca II and in Na I. After removing the stellar continuum using the best-fit model obtained with Prospector, we fit the excess absorption due to neutral gas in the Ca II H, K doublet and in the Na I D doublet, together with nearby emission lines produced by ionized gas. We measure covering fractions between 0.2 and 0.9 from the Ca II H and K lines, which are spectrally well resolved in the NIRSpec R ~ 1000 observations, unlike the absorption lines in the Na I D doublet. We measure the velocity shift, velocity dispersion, and column density separately for Ca II and Na I. About half of the galaxies present blueshifted Ca II, indicative of an outflow of neutral gas, consistent with previous results based on Na I. The velocity shift and the column density measured from Ca II are correlated with those measured from Na I, implying that these absorption lines trace gas in similar physical conditions. However, the column densities are not in a 1:1 relation, meaning that the relative amount of Ca II and Na I atoms along the line of sight varies with the gas column density. After discussing possible reasons for this behavior, we derive an empirical relation between the column density of Ca II and the column density of Na I and, in a more indirect way, of neutral hydrogen H I. This calibration offers a new way to estimate the outflow mass and the mass outflow rate for the neutral phase from current and future JWST observations of massive galaxies at Cosmic Noon and beyond

astro-ph.GA↗

arXiv2026

Empirical Calibration of Na I D and Other Absorption Lines as Tracers of High-Redshift Neutral Outflows

Recent JWST observations of massive galaxies at z > 2 have detected blueshifted absorption in Na I D and other resonant absorption lines, indicative of strong gas outflows in the neutral phase. However, the measured mass outflow rates are highly uncertain because JWST observations can only probe the column density of trace elements such as sodium, while most of the gas is in the form of hydrogen. The conversion between the column density of sodium and that of hydrogen is based on observations of gas clouds within the Milky Way, and has not been directly tested for massive galaxies at high redshift. In order to test this conversion, we study a unique system consisting of a massive quiescent galaxy (J1439B) at z = 2.4189 located at a projected distance of 38 physical kpc from the bright background quasar QSO J1439. The neutral outflow from the galaxy is observed as a sub-damped Lyman-alpha absorber in the spectrum of the background quasar, which enables a direct measurement of the hydrogen column density from Lyman transitions. We obtain new near-infrared spectroscopy with Magellan/FIRE and detect Na I D and other resonant absorption lines from Mg II, Mg I, and Fe II. We are thus able to derive new, empirical calibrations between the column density of trace elements and the hydrogen column density, that can be used to estimate the mass and the rate of neutral gas outflows in other massive quiescent galaxies at high redshift. The calibration we derive for Na I is only 30% lower than the local relation that is typically assumed at high redshift, confirming that the neutral outflows observed with JWST at z > 2 are able to remove a large amount of gas and are thus likely to play a key role in galaxy quenching. However, using the local calibration for Mg II yields an order-of-magnitude discrepancy compared to the empirical calibration, possibly because of variations in the dust depletion.

astro-ph.GA↗

arXiv2026

Human-Centered LLM-Agent System for Detecting Anomalous Digital Asset Transactions

We present HCLA, a human-centered multi-agent system for anomaly detection in digital-asset transactions. The system integrates three cognitively aligned roles: Rule Abstraction, Evidence Scoring, and Expert-Style Justification. These roles operate in a conversational workflow that enables non-experts to express analytical intent in natural language, inspect structured risk evidence, and obtain traceable, context-aware reasoning. Implemented with an open-source, web-based interface, HCLA translates user intent into explicit analytical rules, applies classical anomaly detectors to quantify evidential risk, and reconstructs expert-style justifications grounded in observable transactional signals. Experiments on a cryptocurrency anomaly dataset show that, while the underlying detector achieves strong predictive accuracy, HCLA substantially improves interpretability, interaction, and decision transparency. Importantly, HCLA is not designed to explain a black-box model in the conventional XAI sense. Instead, we reconstruct a traceable expert reasoning process that aligns algorithmic evidence with regulatory and investigative judgment. By explicitly separating evidence scoring from expert-style justification, the framework emphasizes accountability beyond explainability and addresses practical requirements for regulatory, audit, and compliance-driven financial forensics. We describe the system architecture, closed-loop interaction design, datasets, evaluation protocol, and limitations. We argue that a human-in-the-loop reasoning reconstruction paradigm is essential for achieving transparency, accountability, and trust in high-stakes financial environments. Keywords: Human-Centered AI; LLM-Agent System; Multi-Agent Architecture; Anomaly Detection; Digital Asset Transactions; Cryptocurrency Forensics; Blockchain Analytics; Human-in-the-Loop; Explainable AI (XAI); Interpretability

cs.AI↗

arXiv2026

Knowledge-Integrated Representation Learning for Crypto Anomaly Detection under Extreme Label Scarcity; Relational Domain-Logic Integration with Retrieval-Grounded Context and Path-Level Explanations

Detecting anomalous trajectories in decentralized crypto networks is fundamentally challenged by extreme label scarcity and the adaptive evasion strategies of illicit actors. While Graph Neural Networks (GNNs) effectively capture local structural patterns, they struggle to internalize multi hop, logic driven motifs such as fund dispersal and layering that characterize sophisticated money laundering, limiting their forensic accountability under regulations like the FATF Travel Rule. To address this limitation, we propose Relational Domain Logic Integration (RDLI), a framework that embeds expert derived heuristics as differentiable, logic aware latent signals within representation learning. Unlike static rule based approaches, RDLI enables the detection of complex transactional flows that evade standard message passing. To further account for market volatility, we incorporate a Retrieval Grounded Context (RGC) module that conditions anomaly scoring on regulatory and macroeconomic context, mitigating false positives caused by benign regime shifts. Under extreme label scarcity (0.01%), RDLI outperforms state of the art GNN baselines by 28.9% in F1 score. A micro expert user study further confirms that RDLI path level explanations significantly improve trustworthiness, perceived usefulness, and clarity compared to existing methods, highlighting the importance of integrating domain logic with contextual grounding for both accuracy and explainability.

cs.LG↗

arXiv2025

The Blue Jay Survey: Deep JWST Spectroscopy for a Representative Sample of Galaxies at Cosmic Noon

We present the Blue Jay survey, a Cycle-1 JWST program aimed at studying the stellar and gas content of galaxies at Cosmic Noon. The survey consists of deep spectroscopy for 153 targets observed over two pointings in the COSMOS field using the NIRSpec micro-shutter assembly (MSA). We employ the three medium-resolution gratings G140M, G235M, and G395M, with exposure times of 13 hours, 3.2 hours, and 1.6 hours, respectively. We thus obtain full coverage of the 1-5 micron range, corresponding to the entire rest-frame optical wavelength range. The sample is carefully selected to provide a census of galaxies over the redshift range 1.7 < z < 3.5 above a redshift-dependent minimum stellar mass that ranges from 10^8.7 Msun to 10^9.3 Msun.The Blue Jay sample is representative of the entire galaxy population at these redshifts, without strong biases in color, star formation rate, or other properties. The sizes of massive galaxies at these redshifts are comparable to the NIRSpec shutters, which requires custom strategies for designing and reducing the observations. Since the standard A-B nod subtraction leads to flux self-subtraction, we construct a master background from empty shutters and subtract it from each of the science spectra. This, in turn, allows for the use of shorter slitlets consisting of only two shutters per galaxy instead of the usual three, with a substantial increase in the multiplexing of the NIRSpec MSA. We measure multi-band photometry using archival JWST and HST observations in two different ways: in a large elliptical aperture encompassing the entire source and from the exact area in the sky where the NIRSpec 1D spectrum is extracted. This enables self-consistent fits of spectroscopic and photometric data. The Blue Jay dataset, which we publicly release, represents the ideal sample for studying the stellar populations, neutral gas, and ionized gas in Cosmic Noon galaxies.

astro-ph.GA↗

arXiv2025

Improving Cryptocurrency Pump-and-Dump Detection through Ensemble-Based Models and Synthetic Oversampling Techniques

This study aims to detect pump and dump (P&D) manipulation in cryptocurrency markets, where the scarcity of such events causes severe class imbalance and hinders accurate detection. To address this issue, the Synthetic Minority Oversampling Technique (SMOTE) was applied, and advanced ensemble learning models were evaluated to distinguish manipulative trading behavior from normal market activity. The experimental results show that applying SMOTE greatly enhanced the ability of all models to detect P&D events by increasing recall and improving the overall balance between precision and recall. In particular, XGBoost and LightGBM achieved high recall rates (94.87% and 93.59%, respectively) with strong F1-scores and demonstrated fast computational performance, making them suitable for near real time surveillance. These findings indicate that integrating data balancing techniques with ensemble methods significantly improves the early detection of manipulative activities, contributing to a fairer, more transparent, and more stable cryptocurrency market.

cs.AI↗

arXiv2025

MESA Isochrones and Stellar Tracks (MIST) III. The White Dwarf Cooling Sequence

We present a substantial update to the MESA Isochrones and Stellar Tracks (MIST) library, extending the MIST model grids and isochrones down the white dwarf (WD) cooling sequence with realistic physics for WD cooling timescales. This work provides a large grid of MESA models for carbon-oxygen core WDs with hydrogen atmospheres (spectral type DA/DC), descended from full prior stellar evolution calculations. The model tracks, isochrones, and WD cooling timescale contours are available on the MIST project website and at https://doi.org/10.5281/zenodo.15242046. Our WD models provide a very large, publicly available grid with detailed physics for WD cooling timescales: realistic interior and envelope compositions, with element diffusion and heavy-element sedimentation, nuclear burning at the base of the WD hydrogen envelope, core crystallization, and C/O phase separation. As a large grid of open-source stellar evolution models, these WD models provide both out-of-the-box model tracks for comparison with observations and a framework for building further WD models to investigate variations in WD physics.

astro-ph.SR↗

arXiv2025

Hybrid GCN-GRU Model for Anomaly Detection in Cryptocurrency Transactions

Blockchain transaction networks are complex, with evolving temporal patterns and inter-node relationships. To detect illicit activities, we propose a hybrid GCN-GRU model that captures both structural and sequential features. Using real Bitcoin transaction data (2020-2024), our model achieved 0.9470 Accuracy and 0.9807 AUC-ROC, outperforming all baselines.

cs.LG↗

arXiv2025

HyPV-LEAD: Proactive Early-Warning of Cryptocurrency Anomalies through Data-Driven Structural-Temporal Modeling

Abnormal cryptocurrency transactions - such as mixing services, fraudulent transfers, and pump-and-dump operations -- pose escalating risks to financial integrity but remain notoriously difficult to detect due to class imbalance, temporal volatility, and complex network dependencies. Existing approaches are predominantly model-centric and post hoc, flagging anomalies only after they occur and thus offering limited preventive value. This paper introduces HyPV-LEAD (Hyperbolic Peak-Valley Lead-time Enabled Anomaly Detection), a data-driven early-warning framework that explicitly incorporates lead time into anomaly detection. Unlike prior methods, HyPV-LEAD integrates three innovations: (1) window-horizon modeling to guarantee actionable lead-time alerts, (2) Peak-Valley (PV) sampling to mitigate class imbalance while preserving temporal continuity, and (3) hyperbolic embedding to capture the hierarchical and scale-free properties of blockchain transaction networks. Empirical evaluation on large-scale Bitcoin transaction data demonstrates that HyPV-LEAD consistently outperforms state-of-the-art baselines, achieving a PR-AUC of 0.9624 with significant gains in precision and recall. Ablation studies further confirm that each component - PV sampling, hyperbolic embedding, and structural-temporal modeling - provides complementary benefits, with the full framework delivering the highest performance. By shifting anomaly detection from reactive classification to proactive early-warning, HyPV-LEAD establishes a robust foundation for real-time risk management, anti-money laundering (AML) compliance, and financial security in dynamic blockchain environments.

cs.LG↗

arXiv2025

Making the Right Thing: Bridging HCI and Responsible AI in Early-Stage AI Concept Selection

AI projects often fail due to financial, technical, ethical, or user acceptance challenges -- failures frequently rooted in early-stage decisions. While HCI and Responsible AI (RAI) research emphasize this, practical approaches for identifying promising concepts early remain limited. Drawing on Research through Design, this paper investigates how early-stage AI concept sorting in commercial settings can reflect RAI principles. Through three design experiments -- including a probe study with industry practitioners -- we explored methods for evaluating risks and benefits using multidisciplinary collaboration. Participants demonstrated strong receptivity to addressing RAI concerns early in the process and effectively identified low-risk, high-benefit AI concepts. Our findings highlight the potential of a design-led approach to embed ethical and service design thinking at the front end of AI innovation. By examining how practitioners reason about AI concepts, our study invites HCI and RAI communities to see early-stage innovation as a critical space for engaging ethical and commercial considerations together.

cs.HC↗

arXiv2025

Exploring the Innovation Opportunities for Pre-trained Models

Innovators transform the world by understanding where services are successfully meeting customers' needs and then using this knowledge to identify failsafe opportunities for innovation. Pre-trained models have changed the AI innovation landscape, making it faster and easier to create new AI products and services. Understanding where pre-trained models are successful is critical for supporting AI innovation. Unfortunately, the hype cycle surrounding pre-trained models makes it hard to know where AI can really be successful. To address this, we investigated pre-trained model applications developed by HCI researchers as a proxy for commercially successful applications. The research applications demonstrate technical capabilities, address real user needs, and avoid ethical challenges. Using an artifact analysis approach, we categorized capabilities, opportunity domains, data types, and emerging interaction design patterns, uncovering some of the opportunity space for innovation with pre-trained models.

cs.HC↗

arXiv2025

Big, Dusty Galaxies in Blue Jay: Insights into the Relationship Between Morphology and Dust Attenuation at Cosmic Noon

The dust attenuation of galaxies is highly diverse and closely linked to stellar population properties and the star dust geometry, yet its relationship to galaxy morphology remains poorly understood. We present a study of 141 galaxies ($9<\log(\rm M_{\star}/\rm M_{\odot})<11.5$) at $1.7<z<3.5$ from the Blue Jay survey combining deep JWST/NIRCam imaging and $R\sim1000$ JWST/NIRSpec spectra. Using \texttt{Prospector} to perform a joint analysis of these data with non-parametric star-formation histories and a two-component dust model with flexible attenuation laws, we constrain stellar and nebular properties. We find that the shape and strength of the attenuation law vary systematically with optical dust attenuation ($A_V$), stellar mass, and star formation rate (SFR). $A_V$ correlates strongly with stellar mass for starbursts, star-forming galaxies and quiescent galaxies. The inclusion of morphological information tightens these correlations: attenuation correlates more strongly with stellar mass and SFR surface densities than with the global quantities. The Balmer decrement-derived nebular attenuation for 67 of these galaxies shows consistent trends with the stellar continuum attenuation. We detect a wavelength-dependent size gradient: massive galaxies ($\rm M_{\star}\gtrsim 10^{10}~M_{\odot}$) appear $\sim30\%$ larger in the rest-optical than in the rest-NIR, driven by central dust attenuation that flattens optical light profiles. Lower-mass systems exhibit more diverse size ratios, consistent with either inside-out growth or central starbursts. These results demonstrate that dust attenuation significantly alters observed galaxy structure and highlight the necessity of flexible attenuation models for accurate physical and morphological inference at cosmic noon.

astro-ph.GA↗

arXiv2025

AGN Feedback in Quiescent Galaxies at Cosmic Noon Traced by Ionized Gas Emission

We analyze ionized gas emission lines in deep rest-frame optical spectra of 16 quiescent galaxies at redshift $1.7<z<3.5$ observed with JWST/NIRSpec by the Blue Jay survey. Robust detection of emission lines in $75\%$ of the sample indicates the presence of ongoing ionizing sources in this passive population. The H$α$ line luminosities confirm that the population is quiescent, with star formation rates that are at least ten times lower than the main sequence of star formation. The quiescent sample is clearly separate from the star-forming population in line diagnostic diagrams, and occupies a region usually populated by active galactic nuclei (AGN). Analysis of the observed line ratios, equivalent widths, and velocity dispersions leads us to conclude that in most cases the gas is ionized by AGN activity, despite the lack of X-ray detections. A subset of the sample also hosts ionized and/or neutral outflows. Our results show, for the first time using a representative sample, that low luminosity AGN are extremely common among quiescent galaxies at high redshift. These low luminosity AGN may play a key role in quenching star formation and in maintaining massive galaxies quiescent from Cosmic Noon to $z\sim0$.

astro-ph.GA↗

arXiv2025

Discrepancies Between JWST Observations and Simulations of Quenched Massive Galaxies at $z > 3$: A Comparative Study With IllustrisTNG and ASTRID

Recent JWST observations have uncovered an unexpectedly large population of massive quiescent galaxies at $z>3$. Using the cosmological simulations IllustrisTNG and ASTRID, we identify analogous galaxies and investigate their abundance, formation, quenching mechanisms, and post-quenching evolution for stellar masses $9.5 < \log_{10}{(M_\star/{\rm M}_\odot)} < 12$. We apply three different quenching definitions and find that both simulations significantly underestimate the comoving number density of quenched massive galaxies at $z \gtrsim 3$ compared to JWST observations by up to $\sim 2$ dex. In both simulations, the high-$z$ quenched massive galaxies often host overmassive central black holes above the local $M_{BH}-M_\star$ relation, implying that AGN feedback is key in quenching galaxies in the early Universe. The typical quenching timescales for these galaxies are $\sim 200-600$ Myr. IllustrisTNG primarily employs AGN kinetic feedback, while ASTRID relies on AGN thermal feedback at $z > 2.3$, which is less effective and has a longer quenching timescale. Although these simulations differ in many aspects, making a direct comparison challenging, our findings suggest the need for improved physical models of AGN feedback in galaxy formation simulations. At lower stellar masses, the quenched galaxies have denser local environments than the star-forming galaxies, suggesting that environmental quenching helps quench less massive galaxies. We also study the post-quenching evolution of the high-$z$ massive quiescent galaxies and find that many experience subsequent reactivation of star formation, evolving into primary progenitors of $z=0$ brightest cluster galaxies.

astro-ph.GA↗

arXiv2024

Rapid Quenching of Galaxies at Cosmic Noon

The existence of massive quiescent galaxies at high redshift seems to require rapid quenching, but it is unclear whether all quiescent galaxies have gone through this phase and what physical mechanisms are involved. To study rapid quenching, we use rest-frame colors to select 12 young quiescent galaxies at $z \sim 1.5$. From spectral energy distribution fitting, we find that they all experienced intense starbursts prior to rapid quenching. We confirm this with deep Magellan/FIRE spectroscopic observations for a subset of seven galaxies. Broad emission lines are detected for two galaxies and are most likely caused by AGN activity. The other five galaxies do not show any emission features, suggesting that gas has already been removed or depleted. Most of the rapidly quenched galaxies are more compact than normal quiescent galaxies, providing evidence for a central starburst in the recent past. We estimate an average transition time of $300\,\rm Myr$ for the rapid quenching phase. Approximately $4\%$ of quiescent galaxies at $z=1.5$ have gone through rapid quenching; this fraction increases to $23\%$ at $z=2.2$. We identify analogs in the TNG100 simulation and find that rapid quenching for these galaxies is driven by AGN, and for half of the cases, gas-rich major mergers seem to trigger the starburst. We conclude that these massive quiescent galaxies are not just rapidly quenched but also rapidly formed through a major starburst. We speculate that mergers drive gas inflow towards the central regions and grow supermassive black holes, leading to rapid quenching by AGN feedback.

astro-ph.GA↗

arXiv2024

Widespread rapid quenching at cosmic noon revealed by JWST deep spectroscopy

Massive quiescent galaxies in the young universe are expected to be quenched rapidly, but it is unclear whether they all experience starbursts before quenching and what physical mechanism drives rapid quenching. We study 14 massive quiescent galaxies ($\log(M_\star/M_\odot) > 10$) at $z\sim2$ selected from a representative sample of the Blue Jay survey. We reconstruct their star formation histories by fitting spectral energy distribution models to the JWST/NIRSpec $R\sim1000$ spectra. We find that massive quiescent galaxies can be split into three categories with roughly equal numbers of galaxies according to their SFHs: 1) Relatively old galaxies quenched at early epochs; 2) Galaxies that are rapidly and recently quenched after a flat or bursty formation history (depending on the assumed prior); 3) Galaxies that are rapidly and recently quenched after a major starburst. Most recently quenched galaxies show neutral gas outflows, probed by blueshifted $\rm Na\,I\,D$ absorption, and ionized gas emission, with line ratios consistent with active galactic nucleus (AGN) diagnostics. This suggests that AGN activity drives multi-phase gas outflows, leading to rapid quenching. By tracing back the SFHs of the entire sample, we predict the number density of massive quiescent galaxies at $z=4-6$: $n=(1.5-6.0)\times10^{-5}\,\rm Mpc^{-3}$. The two old massive quiescent galaxies in our sample appear to have extremely early formation and quenching ($z\gtrsim6$), possibly descendants of early post-starbursts at $z>3$. These galaxies still show neutral gas reservoirs and weak H$α$ emission, perhaps because the ejective AGN feedback that caused rapid quenching has weakened over time.

astro-ph.GA↗