arXiv subjects

Xiao Tian

Publications and source records attributed to Xiao Tian.

12 recordsLinked to original sources

arXiv2026

Is Data Shapley Not Better than Random in Data Selection? Ask NASH

Data selection studies the problem of identifying high-quality subsets of training data. While some existing works have considered selecting the subset of data with top-$m$ Data Shapley or other semivalues as they account for the interaction among every subset of data, other works argue that Data Shapley can sometimes perform ineffectively in practice and select subsets that are no better than random. This raises the questions: (I) Are there certain "Shapley-informative" settings where Data Shapley consistently works well? (II) Can we strategically utilize these settings to select high-quality subsets consistently and efficiently? In this paper, we propose a novel data selection framework, NASH (Non-linear Aggregation of SHapley-informative components), which (I) decomposes the target utility function (e.g., validation accuracy) into simpler, Shapley-informative component functions, and selects data by optimizing an objective that (II) aggregates these components non-linearly. We demonstrate that NASH substantially boosts the effectiveness of Shapley/semivalue-based data selection with minimal additional runtime cost.

cs.LG↗

arXiv2026

Incentivizing Truthfulness and Collaborative Fairness in Bayesian Learning

Collaborative machine learning involves training high-quality models using datasets from a number of sources. To incentivize sources to share data, existing data valuation methods fairly reward each source based on its data submitted as is. However, as these methods do not verify nor incentivize data truthfulness, the sources can manipulate their data (e.g., by submitting duplicated or noisy data) to artificially increase their valuations and rewards or prevent others from benefiting. This paper presents the first mechanism that provably ensures (F) collaborative fairness and incentivizes (T) truthfulness at equilibrium for Bayesian models. Our mechanism combines semivalues (e.g., Shapley value), which ensure fairness, and a truthful data valuation function (DVF) based on a validation set that is unknown to the sources. As semivalues are influenced by others' data, we introduce an additional condition to prove that a source can maximize its expected data values in coalitions and semivalues by submitting a dataset that captures its true knowledge. Additionally, we discuss the implications and suitable relaxations of (F) and (T) when the mediator has a limited budget for rewards or lacks a validation set. Our theoretical findings are validated on synthetic and real-world datasets.

cs.LG↗

arXiv2026

INO-SGD: Addressing Utility Imbalance under Individualized Differential Privacy

Differential privacy (DP) is widely employed in machine learning to protect confidential or sensitive training data from being revealed. As data owners gain greater control over their data due to personal data ownership, they are more likely to set their own privacy requirements, necessitating individualized DP (IDP) to fulfil such requests. In particular, owners of data from more sensitive subsets, such as positive cases of stigmatized diseases, likely set stronger privacy requirements, as leakage of such data could incur more serious societal impact. However, existing IDP algorithms induce a critical utility imbalance problem: Data from owners with stronger privacy requirements may be severely underrepresented in the trained model, resulting in poorer performance on similar data from subsequent users during deployment. In this paper, we analyze this problem and propose the INO-SGD algorithm, which strategically down-weights data within each batch to improve performance on the more private data across all iterations. Notably, our algorithm is specially designed to satisfy IDP, while existing techniques addressing utility imbalance neither satisfy IDP nor can be easily adapted to do so. Lastly, we demonstrate the empirical feasibility of our approach.

cs.LG↗

arXiv2026

Short-duration GRB 250221A Afterglow Driven by Two-Component Jets from the merger of a compact star

GRB 250221A is a short gamma-ray burst (GRB) at redshift $z=0.768$, with a duration of 1.8 s and no extended emission in either Swift/Burst Alert Telescope or Konus-Wind bands. A remarkable rebrightening feature in both optical and X-ray bands was observed at $\sim$0.6 days after the burst trigger, but no supernova or kilonova signature was detected. The burst properties and empirical correlations or distributions (e.g., duration, spectral hardness, location in the Amati correlation, $\varepsilon-$value, $f_{\rm eff}$ parameter, and physical offset) favor a compact binary merger origin. However, a dense circumburst medium with $n\sim 80\rm~cm^{-3}$, obtained by adopting the energy injection into a jet to interpret the late-time rebrightening is inconsistent with the compact binary merger origin. In this paper, we propose a two-component jet model to explain the multiwavelength afterglow observations of GRB 250221A, in which the relativistic narrow jet ($\rm θ_{c} \sim 3.8^\circ$) produces the prompt and the early decay afterglow emission, while the mildly relativistic wide jet ($\rm θ_{w} \sim 4.4^\circ$) dominates at later times, resulting in the observed rebrightening feature. If this is the case, one can obtain a lower medium density with $n\sim 0.72\rm~cm^{-3}$ which is a little bit higher than that of short GRBs in merger environments, but falls into the reasonable and acceptable range. Finally, a possible kilonova emission is also discussed within the scenario of compact star merger origin of GRB 250221A.

astro-ph.HE↗

arXiv2025

Probing the Nature of High-Redshift Long GRB 250114A and Its Magnetar Central Engine

GRB 250114A is a long-duration gamma-ray burst (GRB) which triggered the Swift/BAT with a spectroscopic high-redshift at $z = 4.732$. The light curve of the prompt emission is composed of three distinct emission episodes, which are separated by quiescent gaps ranging from tens to hundreds of seconds. While the X-ray light curve exhibits the canonical X-ray emission which is composed of several power-law segments superposition of a giant X-ray flare. More interestingly, there is still significant X-ray emission during the quiescent time in the prompt emission, suggesting a continuously active central engine whose power fluctuates across the $γ$-ray detectability threshold. In this paper, we propose a magnetar as the central engine of GRB 250114A by fitting the X-ray light curve, and infer a magnetic field strength $B_{\rm p}=13.24^{+1.73}_{-5.84} \, \times10^{15}\ \mathrm{G}$ and an initial spin period $P_{0}=14.31^{+0.93}_{-3.16} \, \mathrm{ms}$ of magnetar, with a jet correction, fall within a reasonable range. Furthermore, we also compare the prompt emission, X-ray afterglow, $E_{\mathrm p}$-$E_{γ,\mathrm{iso}}$, and $\varepsilon-$distribution of GBR 250114A with those of other high-$z$ sample-GRBs, and find no significant statistical differences between them.

astro-ph.HE↗

arXiv2025

Signature of strange star as the central engine of GRB 240529A

GRB 240529A is a long-duration gamma-ray burst (GRB) whose light curve of prompt emission is composed of a triple-episode structure, separated by quiescent gaps of tens to hundreds of seconds. More interestingly, its X-ray light curve of afterglow exhibits two-plateau emissions, namely, an internal plateau emission that is smoothly connected with a $\sim t^{-0.1}$ segment and followed by a $\sim t^{-2}$ power-law decay. The three episodes in the prompt emission, together with two plateau emissions in X-ray, are unique in the Swift era. They are very difficult to explain with the standard internal/external shock model by invoking a black hole central engine. However, it could be consistent with the prediction of a supramassive magnetar as the central engine, the physical process of phase transition from magnetar to strange star, as well as the cooling and spin-down of the strange star. In this paper, we propose that the first- and second-episode emissions in the prompt $γ-$ray of GRB 240529A are from the jet emission of a massive star collapsing into a supramassive magnetar and the re-activity of central engine, respectively. Then, the third-episode emission of prompt is attributed to the phase transition from a magnetar to a strange star. Finally, the first- and second-plateau emissions of the X-ray afterglow are powered by the cooling and spin-down of the strange star, respectively. The observational data of each component of GRB 240529A are roughly coincident with the estimations of the above physical picture.

astro-ph.HE↗

arXiv2024

The Progenitor and Central Engine of short-duration GRB 201006A associated with a coherent radio flash

Recently, the detection of a coherent radio flash associated with short-duration GRB 201006A, occurring 76.6 minutes after the burst, has attracted great attention. However, the physical origin of the coherent radio flash remains under debate. By reanalyzing its data observed by Fermi and Swift, we find that an early radio afterglow as the physical origin of the radio flash can be ruled out, but the coherent radio emission seems to be consistent with the hypothesis of a supramassive magnetar as the central engine collapsing into a black hole. Within this scenario, the derived magnetar surface magnetic field ($B_{\rm p}$) and the initial spin period ($P_{\rm 0}$) fall into a reasonable range but require a preferable low value of $η_{\rm R} = 10^{-7}$ or $10^{-6}$. Moreover, the calculated low-$\varepsilon$ value and $E_{\rm γ,iso}-E_{\rm p}$ correlation of GRB 201006A also supports the progenitor which is from the merger of compact stars. We also discuss the non-detected kilonova emission associated with GRB 201006A, and then compare with its upper limits of optical observations.

astro-ph.HE↗

arXiv2024

DeRDaVa: Deletion-Robust Data Valuation for Machine Learning

Data valuation is concerned with determining a fair valuation of data from data sources to compensate them or to identify training examples that are the most or least useful for predictions. With the rising interest in personal data ownership and data protection regulations, model owners will likely have to fulfil more data deletion requests. This raises issues that have not been addressed by existing works: Are the data valuation scores still fair with deletions? Must the scores be expensively recomputed? The answer is no. To avoid recomputations, we propose using our data valuation framework DeRDaVa upfront for valuing each data source's contribution to preserving robust model performance after anticipated data deletions. DeRDaVa can be efficiently approximated and will assign higher values to data that are more useful or less likely to be deleted. We further generalize DeRDaVa to Risk-DeRDaVa to cater to risk-averse/seeking model owners who are concerned with the worst/best-cases model utility. We also empirically demonstrate the practicality of our solutions.

cs.LG↗

arXiv2023

Radio Plateaus in Gamma-Ray Burst Afterglows and Their Application in Cosmology

The plateau phase in the radio afterglows has been observed in very few gamma-ray bursts (GRBs), and 27 radio light curves with plateau phase were acquired from the published literature in this article. We obtain the related parameters of the radio plateau, such as temporal indexes during the plateau phase ($α_1$ and $α_2$), break time ($\Tbz$) and the corresponding radio flux ($F_{\rm b}$). The two parameter Dainotti relation between the break time of the plateau and the corresponding break luminosity ($\Lbz$) in radio band is $\Lbz \propto \Tbz^{-1.20\pm0.24}$. Including the isotropic energy $\Eiso$ and the peak energy $\Epi$, the three parameter correlations for the radio plateaus are written as $\Lbz \propto \Tbz^{-1.01 \pm 0.24} \Eiso^{0.18 \pm 0.09}$ and $\Lbz \propto \Tbz^{-1.18 \pm 0.27} \Epi^{0.05 \pm 0.28}$, respectively. The correlations are less consistent with that of X-ray and optical plateaus, implying that radio plateaus may have a different physical mechanism. The typical frequencies crossing the observational band may be a reasonable hypothesis that causes the breaks of the radio afterglows. We calibrate GRBs empirical luminosity correlations as standard candle for constraining cosmological parameters, and find that our samples can constrain the flat $Λ$CDM model well, while are not sensitive to non-flat $Λ$CDM model. By combining GRBs with other probes, such as SN and CMB, the constraints on cosmological parameters are $\om = 0.297\pm0.006$ for the flat $Λ$CDM model and $\om = 0.283\pm0.008$, $\oL = 0.711\pm0.006$ for the non-flat $Λ$CDM model, respectively.

astro-ph.HE↗

arXiv2021

Constraining the circumburst medium of gamma-ray bursts with X-ray afterglows

Long gamma-ray bursts (GRBs) are considered to be originated from core collapse of massive stars. It is believed that the afterglow property is determined by the density of the material in the surrounding interstellar medium. Therefore, the circumburst density can be used to distinguish between an interstellar wind $n(R) \propto R^{\rm -k}$, and a constant density medium (ISM), $n(R) = const$. Previous studies with different afterglow samples, show that the circumburst medium of GRBs is neither simply supported by an interstellar wind, nor completely favored by an ISM. In this work, our new sample is consisted of 39 GRBs with smoothly onset bump-like features in early X-ray afterglows, in which 20 GRBs have the redshift measurements. By using a smooth broken power law function to fit the bumps of X-ray light curves, we derive the full width at half-maximum (FWHM) as the feature width ($ω$), as well as the rise and decay time scales of the bumps ($T_{\rm r}$ and $T_{\rm d}$). The correlations between the timescales of X-ray bumps are similar to those found previously in the optical afterglows. Based on the fireball forward shock (FS) model of the thin shell case, we obtain the distribution of the electron spectral index $p$, and further constrain the medium density distribution index $k$. The new inferred $k$ is found to be concentrated at 1.0, with a range from 0.2 to 1.8. This finding is consistent with previous studies. The conclusion of our detailed investigation for X-ray afterglows suggests that the ambient medium of the selected GRBs is not homogeneous, i.e., neither ISM nor the typical interstellar wind.

astro-ph.HE↗

arXiv2020

Real-time Earthquake Early Warning with Deep Learning: Application to the 2016 Central Apennines, Italy Earthquake Sequence

Earthquake early warning systems are required to report earthquake locations and magnitudes as quickly as possible before the damaging S wave arrival to mitigate seismic hazards. Deep learning techniques provide potential for extracting earthquake source information from full seismic waveforms instead of seismic phase picks. We developed a novel deep learning earthquake early warning system that utilizes fully convolutional networks to simultaneously detect earthquakes and estimate their source parameters from continuous seismic waveform streams. The system determines earthquake location and magnitude as soon as one station receives earthquake signals and evolutionarily improves the solutions by receiving continuous data. We apply the system to the 2016 Mw 6.0 earthquake in Central Apennines, Italy and its subsequent sequence. Earthquake locations and magnitudes can be reliably determined as early as four seconds after the earliest P phase, with mean error ranges of 6.8-3.7 km and 0.31-0.23, respectively.

physics.geo-ph↗

arXiv2019

Identifying the Geometry of an Object Using Lock-In Thermography

Lock-in Thermography (LIT) is a type of Infrared Thermography (IRT) that can be used as a useful non-destructive testing (NDT) technique for the detection of subsurface anomalies in objects. Currently, LIT fails to estimate the thickness at a point on the tested object. This makes LIT unable to figure out the 3-dimensional geometry of an object. In this project, two techniques of identifying the geometry of an object using LIT are discussed. The main idea of both techniques is to find a relationship between the parameters obtained from LIT and the thickness at each data point. Technique I builds a numerical function that models the relationship between thickness, Lock-In phase, and other parameters. The function is then inverted for thickness estimation. Technique II is a quantitative method, in which a database is created with six dimensions - thickness, Lock-In phase, Lock-In amplitude and three other parameters, based on data obtained from LIT experiments or simulations. Estimated thickness is obtained by retrieving data from the database. The database can be improved based on Principal Component Analysis. Evaluation of the techniques is done by measuring root-mean-square deviation, and calculating successful rate with different tolerances. Moreover, during the application of the techniques, Stochastic Gradient Descent can be used to determine the time when sufficient data have been collected from LIT measurement to generate the estimated geometry accurately.

physics.app-ph↗