arXiv subjects

Hooshang Nayyeri

Publications and source records attributed to Hooshang Nayyeri.

At least 19 recordsLinked to original sources

arXiv2026

Learning to See Sharper: A Physics-Informed Artificial Intelligence Framework for Super-Resolving Galaxy Spectra

The information recoverable from galaxy spectra depends fundamentally on spectral resolution, yet assembling large samples at high resolution remains observationally expensive. We present a deep-learning framework for spectral super-resolution that enhances low-resolution galaxy spectra by a factor of $\sim$10 in resolving power ($R\sim100$ to $R\sim1000$). The model is trained on 2,858 paired JWST/NIRSpec observations from the JADES program, where low-resolution prism spectra are matched with medium-resolution grating spectra (G140M, G235M, G395M) combined into a unified reference covering $1.0-5.3μ\mathrm{m}$. Our three-stage architecture performs an initial super-resolution, infers the redshift from the coarse reconstruction, and then applies a physics-informed residual refinement that uses attention across emission-line tokens to learn inter-line relationships and predict parametric line profiles, alongside a convolutional branch for continuum corrections. Evaluated on a 20% held-out sample, the model achieves noise-limited residuals over most of the spectral range and improves the signal-to-noise ratio of key diagnostic lines including [OII], H$β$, [OIII], and H$α$. The super-resolved spectra separate features that are entirely unresolved at prism resolution, such as the [OIII] $λ\lambda4959,5007$ doublet and H$β$, in exchange for a modest retrieval of line fluxes. As a proof of concept using JWST data, this approach is readily extensible to the low-resolution grism spectroscopy that will be delivered by Euclid and the Roman Space Telescope, potentially enabling population-level diagnostics across millions of galaxy spectra that would otherwise be inaccessible at low-resolution.

astro-ph.GA↗

arXiv2026

PersEvalBench: Can LLM Evaluators Judge Contextually Appropriate Use of Persona Information?

Large language model (LLM) evaluators are increasingly used to assess personalized assistants, yet whether they can reliably judge personalization has rarely been tested. We introduce PersEvalBench, a benchmark testing whether an evaluator can recognize appropriate and inappropriate uses of user history. It contains 403 multi-episode scenarios where user information should persist, transfer, or cease to apply. The benchmark covers nine controlled error subtypes across three families: under-personalization, over-personalization, and misapplied-state errors. Human audits confirm that generated errors match their intended types and that the prescribed uses of user information are reasonable. Across five LLM evaluators, recognition of controlled personalization errors remains limited even with the full dialogue history. On the human-evaluated subset, under-personalization shows the largest gap: humans recognize 94.4% of these errors, compared with 63.9% for the strongest reported evaluator. Evaluators often recognize that a response is problematic but reverse the error direction, mistaking failure to follow an applicable preference (UNDER) for use of an inapplicable preference (OVER). Easier evidence access offers little help: highlighting the relevant history turns raises error recognition only slightly, suggesting that the remaining difficulty lies in judging what role that information should play in the response.

cs.CL↗

arXiv2026

ATOD: An Evaluation Framework and Benchmark for Agentic Task-Oriented Dialogue Systems

Agentic task-oriented dialogue (TOD) requires systems to track concurrent goals, dependencies, and long-horizon state. We examine goal-lifecycle recovery from fixed dialogue trajectories. ATOD contains 1,000 synthetic dialogues annotated for six advanced-TOD properties, and ATOD-Eval defines metrics for dependency-sensitive completion, memory recall, and proactivity. We implement a symbolic-vector memory evaluator for ATOD-Eval. Among five prompt-only predictors and six matched-backbone memory baselines, our evaluator is the only configuration above 90% in both goal detection F1 and conditional status accuracy on medium dialogues. On complex dialogues, no baseline exceeds it on both metrics; it has the highest conditional status accuracy within the memory-based block and the lowest measured per-turn latency. These experiments assess lifecycle tracking rather than interactive agent task success. A cross-family judge swap and a manual audit with 94.0% agreement provide initial checks on measurement reliability. Code and data will be released at https://github.com/amazon-science/ATOD.

cs.CL↗

arXiv2026

Hear2Act: Benchmarking When Prosody Should Change What an Assistant Does

Prosodic cues can convey task-relevant information that alters the trajectory and outcome of a task-oriented dialogue, even when the words themselves remain unchanged. Yet existing benchmarks typically evaluate prosodic perception, response appropriateness, and task-oriented dialogue in isolation, making it difficult to test whether prosodic evidence changes downstream decisions. We introduce Hear2Act, a unified evaluation protocol for text and spoken assistants with 480 persona-grounded scenarios, hidden user concerns, and objectively verifiable outcomes. For each scenario, we keep the task and user needs fixed while varying whether the same concern is conveyed explicitly in words or primarily through prosody, and evaluate decisions under transcript, audio, and concern-state access. Using Hear2Act, we evaluate two audio-capable LLMs. Under Prosody-mediated feedback, adding audio to the transcript changes the average optimal-solution rate only from 14.6% to 15.3%. In contrast, when models infer the concern status from audio, represent it in text, and use it for next-action selection, the rate rises to 39.6%, close to 40.7% with the ground-truth state. This contrast, however, largely disappears under Explicit lexical feedback, where the concern is verbally mentioned in the utterance. Together, these results show that prosody matters when lexical evidence is insufficient, and that audio-capable LLMs can recover information from speech but do not reliably carry it into action without an explicit intermediate representation.

cs.CL↗

arXiv2026

Trajectory-Level Redirection Attacks on Vision-Language-Action Models

Vision-language-action (VLA) policies bring natural language into closed-loop robot control, enabling robots to execute manipulation tasks directly from text instructions. The same interface gives text a recurring role in control because the prompt is reused at every replanning step, and each prompt-conditioned action changes the future observations on which the policy acts. Existing VLA attacks study adversarial prompts that elicit targeted low-level actions or make such actions persist across changing images. We identify a stronger trajectory-level failure mode: a prompt that still $\textit{appears}$ to specify the intended task but redirects the final physical outcome. We mathematically formalize this setting as $\textit{command-preserving trajectory redirection}$, a prompt-only threat model in which the attacker chooses one prompt before the episode, all policy and environment components remain fixed, and the prompt must stay close to the benign instruction while omitting target words and correction language. To find such prompts, we introduce an on-policy prompt search method that uses rollouts to discover perturbations whose closed-loop behavior tracks a target task while satisfying the command-preserving constraints. Experiments in simulation and on hardware show that near-benign prompt perturbations can redirect VLA rollouts to attacker-specified targets. These results expose a trajectory-level vulnerability in VLA instruction grounding: text that appears to preserve the intended command can still give an adversary control over the robot's final physical outcome. Project website: https://vla-redirection-attack.github.io/

cs.RO↗

arXiv2026

Beyond Individual Personas: Aligning Synthetic Dialogue to Population-Level Behavior Distributions

Synthetic dialogue corpora are increasingly used as proxies for target dialogue data, yet persona-grounded generators optimize individual conversations rather than corpus composition, yielding locally plausible dialogues with distorted population-level behavior mixes. We introduce GroupPersona, a framework that aligns synthetic dialogue corpora to the behavior distribution of a reference corpus. GroupPersona turns population statistics into generation controls: it separates each dialogue's core behavioral signature from predictable side effects, and uses the resulting behavioral groups to condition user agents on the interaction patterns that define the reference population. We evaluate GroupPersona on four corpora crossing two dialogue sources, assistant-style and Reddit-derived, with two construction variants: structure-preserving and variation-enhanced. GroupPersona lowers Jensen-Shannon divergence between synthetic and reference distributions over 12 behavior attributes from 0.234 to 0.177 relative to the strongest average baseline, a 24.4% reduction, and is best or tied-best on all four corpora while preserving structural alignment. It also achieves the closest calibration to reference-conversation quality scores, reducing mean absolute deviation from the reference-conversation profile to 0.63 versus 0.91 for the next-best baseline.

cs.CL↗

arXiv2026

Interactive World Simulator for Robot Policy Training and Evaluation

Action-conditioned video prediction models (often referred to as world models) have shown strong potential for robotics applications, but existing approaches are often slow and struggle to capture physically consistent interactions over long horizons, limiting their usefulness for scalable robot policy training and evaluation. We present Interactive World Simulator, a framework for building interactive world models from a moderate-sized robot interaction dataset. Our approach leverages consistency models for both image decoding and latent-space dynamics prediction, enabling fast and stable simulation of physical interactions. In our experiments, the learned world models produce interaction-consistent pixel-level predictions and support stable long-horizon interactions for more than 10 minutes at 15 FPS on a single RTX 4090 GPU. Our framework enables scalable demonstration collection solely within the world models to train state-of-the-art imitation policies. Through extensive real-world evaluation across diverse tasks involving rigid objects, deformable objects, object piles, and their interactions, we find that policies trained on world-model-generated data perform comparably to those trained on the same amount of real-world data. Additionally, we evaluate policies both within the world models and in the real world across diverse tasks, and observe a strong correlation between simulated and real-world performance. Together, these results establish the Interactive World Simulator as a stable and physically consistent surrogate for scalable robotic data generation and faithful, reproducible policy evaluation.

cs.RO↗

arXiv2026

MIRAGE: A Benchmark for Multimodal Information-Seeking and Reasoning in Agricultural Expert-Guided Conversations

We introduce MIRAGE, a new benchmark for multimodal expert-level reasoning and decision-making in consultative interaction settings. Designed for the agriculture domain, MIRAGE captures the full complexity of expert consultations by combining natural user queries, expert-authored responses, and image-based context, offering a high-fidelity benchmark for evaluating models on grounded reasoning, clarification strategies, and long-form generation in a real-world, knowledge-intensive domain. Grounded in over 35,000 real user-expert interactions and curated through a carefully designed multi-step pipeline, MIRAGE spans diverse crop health, pest diagnosis, and crop management scenarios. The benchmark includes more than 7,000 unique biological entities, covering plant species, pests, and diseases, making it one of the most taxonomically diverse benchmarks available for vision-language models, grounded in the real world. Unlike existing benchmarks that rely on well-specified user inputs and closed-set taxonomies, MIRAGE features underspecified, context-rich scenarios with open-world settings, requiring models to infer latent knowledge gaps, handle rare entities, and either proactively guide the interaction or respond. Project Page: https://mirage-benchmark.github.io

cs.LG↗

arXiv2025

A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards

Task specification for robotic manipulation in open-world environments is challenging, requiring flexible and adaptive objectives that align with human intentions and can evolve through iterative feedback. We introduce Iterative Keypoint Reward (IKER), a visually grounded, Python-based reward function that serves as a dynamic task specification. Our framework leverages VLMs to generate and refine these reward functions for multi-step manipulation tasks. Given RGB-D observations and free-form language instructions, we sample keypoints in the scene and generate a reward function conditioned on these keypoints. IKER operates on the spatial relationships between keypoints, leveraging commonsense priors about the desired behaviors, and enabling precise SE(3) control. We reconstruct real-world scenes in simulation and use the generated rewards to train reinforcement learning (RL) policies, which are then deployed into the real world-forming a real-to-sim-to-real loop. Our approach demonstrates notable capabilities across diverse scenarios, including both prehensile and non-prehensile tasks, showcasing multi-step task execution, spontaneous error recovery, and on-the-fly strategy adjustments. The results highlight IKER's effectiveness in enabling robots to perform multi-step tasks in dynamic environments through iterative reward shaping.

cs.RO↗

arXiv2024

RoboEXP: Action-Conditioned Scene Graph via Interactive Exploration for Robotic Manipulation

We introduce the novel task of interactive scene exploration, wherein robots autonomously explore environments and produce an action-conditioned scene graph (ACSG) that captures the structure of the underlying environment. The ACSG accounts for both low-level information (geometry and semantics) and high-level information (action-conditioned relationships between different entities) in the scene. To this end, we present the Robotic Exploration (RoboEXP) system, which incorporates the Large Multimodal Model (LMM) and an explicit memory design to enhance our system's capabilities. The robot reasons about what and how to explore an object, accumulating new information through the interaction process and incrementally constructing the ACSG. Leveraging the constructed ACSG, we illustrate the effectiveness and efficiency of our RoboEXP system in facilitating a wide range of real-world manipulation tasks involving rigid, articulated objects, nested objects, and deformable objects.

cs.RO↗

arXiv2024

SPHEREx: NASA's Near-Infrared Spectrophotmetric All-Sky Survey

SPHEREx, the Spectro-Photometer for the History of the Universe, Epoch of Reionization, and ices Explorer, is a NASA MIDEX mission planned for launch in 2024. SPHEREx will carry out the first all-sky spectral survey at wavelengths between 0.75 micron and 5 micron with spectral resolving power ~40 between 0.75 and 3.8 micron and ~120 between 3.8 and 5 micron At the end of its two-year mission, SPHEREx will provide 0.75-to-5 micron spectra of each 6.2"x6.2" pixel on the sky - 14 billion spectra in all. This paper updates an earlier description of SPHEREx presenting changes made during the mission's Preliminary Design Phase, including a discussion of instrument integration and test and a summary of the data processing, analysis, and distribution plans.

astro-ph.IM↗

arXiv2022

The Art of Measuring Physical Parameters in Galaxies: A Critical Assessment of Spectral Energy Distribution Fitting Techniques

The study of galaxy evolution hinges on our ability to interpret multi-wavelength galaxy observations in terms of their physical properties. To do this, we rely on spectral energy distribution (SED) models which allow us to infer physical parameters from spectrophotometric data. In recent years, thanks to the wide and deep multi-waveband galaxy surveys, the volume of high quality data have significantly increased. Alongside the increased data, algorithms performing SED fitting have improved, including better modeling prescriptions, newer templates, and more extensive sampling in wavelength space. We present a comprehensive analysis of different SED fitting codes including their methods and output with the aim of measuring the uncertainties caused by the modeling assumptions. We apply fourteen of the most commonly used SED fitting codes on samples from the CANDELS photometric catalogs at z~1 and z~3. We find agreement on the stellar mass, while we observe some discrepancies in the star formation rate (SFR) and dust attenuation results. To explore the differences and biases among the codes, we explore the impact of the various modeling assumptions as they are set in the codes (e.g., star formation histories, nebular, dust, and AGN models) on the derived stellar masses, SFRs, and A_V values. We then assess the difference among the codes on the SFR-stellar mass relation and we measure the contribution to the uncertainties by the modeling choices (i.e., the modeling uncertainties) in stellar mass (~0.1dex), SFR (~0.3dex), and dust attenuation (~0.3mag). Finally, we present some resources summarizing best practices in SED fitting.

astro-ph.GA↗

arXiv2022

Deblending Galaxies with Generative Adversarial Networks

Deep generative models including generative adversarial networks (GANs) are powerful unsupervised tools in learning the distributions of data sets. Building a simple GAN architecture in PyTorch and training on the CANDELS data set, we generate galaxy images with the Hubble Space Telescope resolution starting from a noise vector. We proceed by modifying the GAN architecture to improve the Subaru Hyper Suprime-Cam ground-based images by increasing their resolution to the HST resolution. We use the super resolution GAN on a large sample of blended galaxies which we create using CANDELS cutouts. In our simulated blend sample, $\sim 20 \%$ would unrecognizably be blended even in the HST resolution cutouts. In the HSC-like cutouts this fraction rises to $\sim 90\%$. With our modified GAN we can lower this value to $\sim 50\%$. We quantify the blending fraction in the high, low and GAN resolutions over the whole manifold of angular separation, flux ratios, sizes and redshift difference between the two blended objects. The two peaks found by the GAN deblender result in ten times improvement in the photometry measurement of the blended objects. Modifying the architecture of the GAN, we also train a Multi-wavelength GAN with seven band optical+NIR HST cutouts. This multi-wavelength GAN improves the fraction of detected blends by another $\sim 10\%$ compared to the single-band GAN. This is most beneficial to the current and future precision cosmology experiments (e.g., LSST, SPHEREx, Euclid, Roman), specifically those relying on weak gravitational lensing, where blending is a major source of systematic error.

astro-ph.CO↗

arXiv2022

Gas Phase Metallicities of Local Ultra-Luminous Infrared Galaxies Follow Normal Star-Forming Galaxies

Despite advances in observational data, theoretical models, and computational techniques to simulate key physical processes in the formation and evolution of galaxies, the stellar mass assembly of galaxies still remains an unsolved problem today. Optical spectroscopic measurements appear to show that the gas-phase metallicities of local ultra-luminous infrared galaxies (ULIRGs) are significantly lower than those of normal star-forming galaxies. This difference has resulted in the claim that ULIRGs are fueled by metal-poor gas accretion from the outskirts\cite{Mannucci10}. Here we report on a new set of gas-phase metallicity measurements making use of the far-infrared spectral lines of [O{\sc iii}]52 $μ$m, [O{\sc iii}]88 $μ$m, and [N{\sc iii}]57 $μ$m instead of the usual optical lines. Photoionization models have resulted in a metallicity diagnostic based on these three lines that break the electron density degeneracy and reduce the scatter of the correlation significantly. Using new data from SOFIA and archival data from Herschel Space Observatory, we find that local ULIRGs lie on the mass-metallicity relation of star-forming galaxies and have metallicities comparable to other galaxies with similar stellar masses and star formation rates. The lack of a departure suggests that ULIRGs follow the same mass assembly mechanism as luminous star-forming galaxies and $\sim 0.3$ dex under-abundance in metallicities derived from optical lines is a result of heavily obscured metal-rich gas which has a negligible effect when using the FIR line diagnostics.

astro-ph.GA↗

arXiv2020

Rise of the Titans: Gas Excitation and Feedback in a Binary Hyper-Luminous Dusty Starburst Galaxy at z~6

We report new observations toward the hyper-luminous dusty starbursting major merger ADFS-27 (z=5.655), using ATCA and ALMA. We detect CO 2-1, 8-7, 9-8, 10-9 and H2O(321-221) emission, and a P-Cygni-shaped OH+(11-01) absorption/emission feature. We also tentatively detect H2O(321-312) and OH+(12-01) emission and CH+(1-0) absorption. We find a total cold molecular mass of M_gas = (2.1+/-0.2) x 10^11 (alpha_CO/1.0) Msun. We also find that the excitation of the star-forming gas is overall moderate for a z>5 dusty starburst, which is consistent with its moderate dust temperature. A high density, high kinetic temperature gas component embedded in the gas reservoir is required to fully explain the CO line ladder. This component is likely associated with the "maximum starburst" nuclei in the two merging galaxies, which are separated by only (140+/-13) km/s along the line of sight and 9.0 kpc in projection. The kinematic structure of both components is consistent with galaxy disks, but this interpretation remains limited by the spatial resolution of the current data. The OH+ features are only detected towards the northern component, which is also the one that is more enshrouded in dust and thus remains undetected up to 1.6 um even in our sensitive new HST/WFC3 imaging. The absorption component of the OH+ line is blueshifted and peaks near the CO and continuum emission peak while the emission is redshifted and peaks offset by 1.7 kpc from the CO and continuum emission peak, suggesting that the gas is associated with a massive molecular outflow from the intensely star-forming nucleus that supplies 125 Msun/yr of enriched gas to its halo.

astro-ph.GA↗

arXiv2020

Investigating the Effect of Galaxy Interactions on AGN Enhancement at $0.5<z<3.0$

Galaxy interactions and mergers are thought to play an important role in the evolution of galaxies. Studies in the nearby universe show a higher AGN fraction in interacting and merging galaxies than their isolated counterparts, indicating that such interactions are important contributors to black hole growth. To investigate the evolution of this role at higher redshifts, we have compiled the largest known sample of major spectroscopic galaxy pairs (2381 with $ΔV <5000$ km s$^{-1}$) at $0.5<z<3.0$ from observations in the COSMOS and CANDELS surveys. We identify X-ray and IR AGN among this kinematic pair sample, a visually identified sample of mergers and interactions, and a mass-, redshift-, and environment-matched control sample for each in order to calculate AGN fractions and the level of AGN enhancement as a function of relative velocity, redshift, and X-ray luminosity. While we see a slight increase in AGN fraction with decreasing projected separation, overall, we find no significant enhancement relative to the control sample at any separation. In the closest projected separation bin ($<25$ kpc, $ΔV <1000$ km s$^{-1}$), we find enhancements of a factor of 0.94$^{+0.21}_{-0.16}$ and 1.00$^{+0.58}_{-0.31}$ for X-ray and IR-selected AGN, respectively. While we conclude that galaxy interactions do not significantly enhance AGN activity on average over $0.5<z<3.0$ at these separations, given the errors and the small sample size at the closest projected separations, our results would be consistent with the presence of low-level AGN enhancement.

astro-ph.GA↗

arXiv2020

The SFR-radius connection: data and implications for wind strength and halo concentration

This paper is one in a series that explores the importance of radius as a second parameter in galaxy evolution. The topic investigated here is the relationship between star formation rate (SFR) and galaxy radius ($R_{\rm e}$) for main-sequence star-forming galaxies. The key observational result is that, over a wide range of stellar mass and redshift in both CANDELS and SDSS, there is little trend between SFR and $R_{\rm e}$ at fixed stellar mass. The Kennicutt-Schmidt law, or any similar density-related star formation law, then implies that smaller galaxies must have lower gas fractions than larger galaxies (at fixed $M_{\ast}$), and this is supported by observations of local star-forming galaxies. We investigate the implication by adopting the equilibrium "bathtub" model: the ISM gas mass is assumed to be constant over time and the net star formation rate is the difference between the accretion rate of gas onto the galaxy from the halo and the outflow rate due to winds. To match the observed null correlation between SFR and radius, the bathtub model requires that smaller galaxies at fixed mass have weaker galactic winds. Our hypothesis is that galaxies are a 2-dimensional family whose properties are set mainly by halo mass and concentration. Galaxy radius and accretion rate plausibly both depend on halo concentration, which predicts how wind strength should vary with $R_{\rm e}$ and SFR.

astro-ph.GA↗

arXiv2020

Spectroscopic Confirmation of a Coma Cluster Progenitor at z ~ 2.2

We report the spectroscopic confirmation of a new protocluster in the COSMOS field at $z$ $\sim$ 2.2, COSMOS Cluster 2.2 (CC2.2), originally identified as an overdensity of narrowband selected H$α$ emitting candidates. With only two masks of Keck/MOSFIRE near-IR spectroscopy in both $H$ ($\sim$ 1.47-1.81 $μ$m) and $K$ ($\sim$ 1.92-2.40 $μ$m) bands ($\sim$ 1.5 hour each), we confirm 35 unique protocluster members with at least two emission lines detected with S/N $>$ 3. Combined with 12 extra members from the zCOSMOS-deep spectroscopic survey (47 in total), we estimate a mean redshift and a line-of-sight velocity dispersion of $z_{mean}$=2.23224 $\pm$ 0.00101 and $σ_{los}$=645 $\pm$ 69 km s$^{-1}$ for this protocluster, respectively. Assuming virialization and spherical symmetry for the system, we estimate a total mass of $M_{vir}$ $\sim$ $(1-2) \times$10$^{14}$ $M_{\odot}$ for the structure. We evaluate a number density enhancement of $δ_{g}$ $\sim$ 7 for this system and we argue that the structure is likely not fully virialized at $z$ $\sim$ 2.2. However, in a spherical collapse model, $δ_{g}$ is expected to grow to a linear matter enhancement of $\sim$ 1.9 by $z$=0, exceeding the collapse threshold of 1.69, and leading to a fully collapsed and virialized Coma-type structure with a total mass of $M_{dyn}$($z$=0) $\sim$ 9.2$\times$10$^{14}$ $M_{\odot}$ by now. This observationally efficient confirmation suggests that large narrowband emission-line galaxy surveys, when combined with ancillary photometric data, can be used to effectively trace the large-scale structure and protoclusters at a time when they are mostly dominated by star-forming galaxies.

astro-ph.GA↗