arXiv subjects

Lin Xiao

Publications and source records attributed to Lin Xiao.

At least 37 records · Page 2Linked to original sources

arXiv2025

Incentivize without Bonus: Provably Efficient Model-based Online Multi-agent RL for Markov Games

Multi-agent reinforcement learning (MARL) lies at the heart of a plethora of applications involving the interaction of a group of agents in a shared unknown environment. A prominent framework for studying MARL is Markov games, with the goal of finding various notions of equilibria in a sample-efficient manner, such as the Nash equilibrium (NE) and the coarse correlated equilibrium (CCE). However, existing sample-efficient approaches either require tailored uncertainty estimation under function approximation, or careful coordination of the players. In this paper, we propose a novel model-based algorithm, called VMG, that incentivizes exploration via biasing the empirical estimate of the model parameters towards those with a higher collective best-response values of all the players when fixing the other players' policies, thus encouraging the policy to deviate from its current equilibrium for more exploration. VMG is oblivious to different forms of function approximation, and permits simultaneous and uncoupled policy updates of all players. Theoretically, we also establish that VMG achieves a near-optimal regret for finding both the NEs of two-player zero-sum Markov games and CCEs of multi-player general-sum Markov games under linear function approximation in an online environment, which nearly match their counterparts with sophisticated uncertainty quantification.

cs.LG↗

arXiv2025

SHYI: Action Support for Contrastive Learning in High-Fidelity Text-to-Image Generation

In this project, we address the issue of infidelity in text-to-image generation, particularly for actions involving multiple objects. For this we build on top of the CONFORM framework which uses Contrastive Learning to improve the accuracy of the generated image for multiple objects. However the depiction of actions which involves multiple different object has still large room for improvement. To improve, we employ semantically hypergraphic contrastive adjacency learning, a comprehension of enhanced contrastive structure and "contrast but link" technique. We further amend Stable Diffusion's understanding of actions by InteractDiffusion. As evaluation metrics we use image-text similarity CLIP and TIFA. In addition, we conducted a user study. Our method shows promising results even with verbs that Stable Diffusion understands mediocrely. We then provide future directions by analyzing the results. Our codebase can be found on polybox under the link: https://polybox.ethz.ch/index.php/s/dJm3SWyRohUrFxn

cs.CV↗

arXiv2025

DiffCL: A Diffusion-Based Contrastive Learning Framework with Semantic Alignment for Multimodal Recommendations

Multimodal recommendation systems integrate diverse multimodal information into the feature representations of both items and users, thereby enabling a more comprehensive modeling of user preferences. However, existing methods are hindered by data sparsity and the inherent noise within multimodal data, which impedes the accurate capture of users' interest preferences. Additionally, discrepancies in the semantic representations of items across different modalities can adversely impact the prediction accuracy of recommendation models. To address these challenges, we introduce a novel diffusion-based contrastive learning framework (DiffCL) for multimodal recommendation. DiffCL employs a diffusion model to generate contrastive views that effectively mitigate the impact of noise during the contrastive learning phase. Furthermore, it improves semantic consistency across modalities by aligning distinct visual and textual semantic information through stable ID embeddings. Finally, the introduction of the Item-Item Graph enhances multimodal feature representations, thereby alleviating the adverse effects of data sparsity on the overall system performance. We conduct extensive experiments on three public datasets, and the results demonstrate the superiority and effectiveness of the DiffCL.

cs.MM↗

arXiv2024

Mass Ratio Distribution of Hierarchical Triple Systems from the LAMOST-MRS Survey

Hierarchical triple-star systems consists of three components organised into an inner binary ($M_{1}$,$M_{2}$) and a more distant outer tertiary ($M_{3}$) star. The LAMOST Medium-Resolution Spectroscopic Survey (LAMOST-MRS) has offered a great sample for the study of triple system populations. We used the Peak Amplitude Ratio (PAR) method to obtain the mass ratio ($q_\mathrm{in}$, $q_\mathrm{out}$) of a triple system from its normalised spectrum. By calculating Cross-Correlation Function (CCF), we determined the correlation between the mass ratio $q_\mathrm{out}$ ($M_{3}$/($M_{1}$+$M_{2}$)) and the amplitude ratio ($A_{3}$/($A_{1}$+$A_{2}$)). We derived $q_\mathrm{in}$ of $0.5-1.0$ and $q_\mathrm{out}$ between 0.2 and 0.8. By fitting a power-law function of the corrected $q_\mathrm{in}$ distribution, the $γ_\mathrm{in}$ are estimated to be $-0.654\pm2.915$, $4.304\pm1.125$ and $11.371\pm1.309$ for A, F and G type stars. The derived $γ_\mathrm{in}$-values increase as the mass decrease, indicating that less massive stars are more likely to have companion stars with similar masses. By fitting a power-law function of the corrected $q_\mathrm{out}$ distribution, the ${γ_\mathrm{out}}$ are estimated to be $-2.016\pm0.172$, $-1.962\pm0.853$ and $-1.238\pm0.141$ for G, F and A type stars, respectively. The ${γ_\mathrm{out}}$-values show a trend of growth toward lower primary star masses.

astro-ph.SR↗

arXiv2024

Multimodal Graph Neural Network for Recommendation with Dynamic De-redundancy and Modality-Guided Feature De-noisy

Graph neural networks (GNNs) have become crucial in multimodal recommendation tasks because of their powerful ability to capture complex relationships between neighboring nodes. However, increasing the number of propagation layers in GNNs can lead to feature redundancy, which may negatively impact the overall recommendation performance. In addition, the existing recommendation task method directly maps the preprocessed multimodal features to the low-dimensional space, which will bring the noise unrelated to user preference, thus affecting the representation ability of the model. To tackle the aforementioned challenges, we propose Multimodal Graph Neural Network for Recommendation (MGNM) with Dynamic De-redundancy and Modality-Guided Feature De-noisy, which is divided into local and global interaction. Initially, in the local interaction process,we integrate a dynamic de-redundancy (DDR) loss function which is achieved by utilizing the product of the feature coefficient matrix and the feature matrix as a penalization factor. It reduces the feature redundancy effects of multimodal and behavioral features caused by the stacking of multiple GNN layers. Subsequently, in the global interaction process, we developed modality-guided global feature purifiers for each modality to alleviate the impact of modality noise. It is a two-fold guiding mechanism eliminating modality features that are irrelevant to user preferences and captures complex relationships within the modality. Experimental results demonstrate that MGNM achieves superior performance on multimodal information denoising and removal of redundant information compared to the state-of-the-art methods.

cs.MM↗

arXiv2024

JWST/MIRI Observations of Newly Formed Dust in the Cold, Dense Shell of the Type IIn SN 2005ip

Dust from core-collapse supernovae (CCSNe), specifically Type IIP SNe, has been suggested to be a significant source of the dust observed in high-redshift galaxies. CCSNe eject large amounts of newly formed heavy elements, which can condense into dust grains in the cooling ejecta. However, infrared (IR) observations of typical CCSNe generally measure dust masses that are too small to account for the dust production needed at high redshifts. Type IIn SNe, classified by their dense circumstellar medium (CSM), are also known to exhibit strong IR emission from warm dust, but the dust origin and heating mechanism have generally remained unconstrained because of limited observational capabilities in the mid-IR. Here, we present a JWST/MIRI Medium Resolution Spectrograph (MRS) spectrum of the Type IIn SN 2005ip nearly 17 years post-explosion. The Type IIn SN 2005ip is one of the longest-lasting and most well-studied SNe observed to date. Combined with a Spitzer mid-IR spectrum of SN 2005ip obtained in 2008, this data set provides a rare 15-year baseline, allowing for a unique investigation of the evolution of dust. The JWST spectrum shows a new high-mass dust component ($\gtrsim0.08$ M$_{\odot}$) that is not present in the earlier Spitzer spectrum. Our analysis shows dust likely formed over the past 15 years in the cold, dense shell (CDS), between the forward and reverse shocks. There is also a smaller mass of carbonaceous dust ($\gtrsim0.005$ M$_{\odot}$) in the ejecta. These observations provide new insights into the role of SN dust production, particularly within the CDS, and its potential contribution to the rapid dust enrichment of the early Universe.

astro-ph.HE↗

arXiv2024

Dual Approximation Policy Optimization

We propose Dual Approximation Policy Optimization (DAPO), a framework that incorporates general function approximation into policy mirror descent methods. In contrast to the popular approach of using the $L_2$-norm to measure function approximation errors, DAPO uses the dual Bregman divergence induced by the mirror map for policy projection. This duality framework has both theoretical and practical implications: not only does it achieve fast linear convergence with general function approximation, but it also includes several well-known practical methods as special cases, immediately providing strong convergence guarantees.

cs.LG↗

arXiv2024

An Adaptive Stochastic Gradient Method with Non-negative Gauss-Newton Stepsizes

We consider the problem of minimizing the average of a large number of smooth but possibly non-convex functions. In the context of most machine learning applications, each loss function is non-negative and thus can be expressed as the composition of a square and its real-valued square root. This reformulation allows us to apply the Gauss-Newton method, or the Levenberg-Marquardt method when adding a quadratic regularization. The resulting algorithm, while being computationally as efficient as the vanilla stochastic gradient method, is highly adaptive and can automatically warmup and decay the effective stepsize while tracking the non-negative loss landscape. We provide a tight convergence analysis, leveraging new techniques, in the stochastic convex and non-convex settings. In particular, in the convex case, the method does not require access to the gradient Lipshitz constant for convergence, and is guaranteed to never diverge. The convergence rates and empirical evaluations compare favorably to the classical (stochastic) gradient method as well as to several other adaptive methods.

math.OC↗

arXiv2024

Joint Linked Component Analysis for Multiview Data

In this work, we propose the joint linked component analysis (joint\_LCA) for multiview data. Unlike classic methods which extract the shared components in a sequential manner, the objective of joint\_LCA is to identify the view-specific loading matrices and the rank of the common latent subspace simultaneously. We formulate a matrix decomposition model where a joint structure and an individual structure are present in each data view, which enables us to arrive at a clean svd representation for the cross covariance between any pair of data views. An objective function with a novel penalty term is then proposed to achieve simultaneous estimation and rank selection. In addition, a refitting procedure is employed as a remedy to reduce the shrinkage bias caused by the penalization.

stat.ML↗

arXiv2024

Sparse and Integrative Principal Component Analysis for Multiview Data

We consider dimension reduction of multiview data, which are emerging in scientific studies. Formulating multiview data as multi-variate data with block structures corresponding to the different views, or views of data, we estimate top eigenvectors from multiview data that have two-fold sparsity, elementwise sparsity and blockwise sparsity. We propose a Fantope-based optimization criterion with multiple penalties to enforce the desired sparsity patterns and a denoising step is employed to handle potential presence of heteroskedastic noise across different data views. An alternating direction method of multipliers (ADMM) algorithm is used for optimization. We derive the l2 convergence of the estimated top eigenvectors and establish their sparsity and support recovery properties. Numerical studies are used to illustrate the proposed method.

stat.ME↗

arXiv2024

The radius variations of accreting main sequence stars and mass transfer instability

Many previous works studied the dynamical timescale mass transfer stability criteria based on the donor response with neglecting the stellar structure of the accretor. In this letter, we investigate the radial response of accretors with mass accumulation and its effect on the binary mass transfer stability. We perform a series of detailed stellar evolution simulations with different types of accretors and obtain the radial variations of stars accreting at different rates. Since the time within which the donor loses half of the original mass has a correlation with the donor mass, we approximately obtain the mean mass transfer rate as a function of mass ratio. Assuming that the common envelope (CE) phase occurs if the accretor radius exceeds the outer Roche lobe radius, we obtain the critical mass ratio of dynamically unstable mass transfer. We find the critical mass ratios for donors filling their Roche lobes at the Main Sequence (MS) and Hertzsprung Gap (HG) stages are smaller than that derived from the radial response of the donor in the traditional way. Our results may suggest that the binary is easier to enter into the CE phase for a donor star at the MS or HG stage than previously believed.

astro-ph.SR↗

arXiv2024

A $λ$ 3 mm line survey towards the circumstellar envelope of the carbon-rich AGB star IRC+10216 (CW Leo)

We present an unbiased $λ$ 3 mm spectral line survey (between 84.5 and 115.8 GHz), conducted by the Purple Mountain Observatory 13.7 meter radio telescope, together with updated modeling results, towards the carbon-rich Asymptotic Giant Branch star, IRC+10216 (CW Leo). A total of 75 spectral lines (96 transitions) are detected, and identified to arise from 19 molecules: C$_2$H, $l$-C$_3$H, C$_4$H, CN, C$_3$N, HC$_3$N, HC$_5$N, HCN, HNC, CH$_3$CN, MgNC, CO, $c$-C$_3$H$_2$, SiC$_2$, SiO, SiS, CS, C$_2$S, C$_3$S, and their isotopologues. Among them, one molecular emission line (H$^{13}$CCCN $J=13-12$) is discovered in IRC+10216 for the first time. The excitation temperature, column density, and fractional abundance of the detected species are deduced by assuming they are in local thermodynamic equilibrium. In addition, the isotopic ratios of [$^{12}$C]/[$^{13}$C], [$^{32}$S]/[$^{34}$S], [$^{28}$Si]/[$^{29}$Si], and [$^{12}$C$^{34}$S]/[$^{13}$C$^{32}$S] are obtained and found to be consistent with previous studies. Finally, we summarize all of the 106 species detected in IRC+10216 to date with their observed and modeled column densities for the convenience of future studies.

astro-ph.SR↗

arXiv2023

Importance Estimation from Multiple Perspectives for Keyphrase Extraction

Keyphrase extraction is a fundamental task in Natural Language Processing, which usually contains two main parts: candidate keyphrase extraction and keyphrase importance estimation. From the view of human understanding documents, we typically measure the importance of phrase according to its syntactic accuracy, information saliency, and concept consistency simultaneously. However, most existing keyphrase extraction approaches only focus on the part of them, which leads to biased results. In this paper, we propose a new approach to estimate the importance of keyphrase from multiple perspectives (called as \textit{KIEMP}) and further improve the performance of keyphrase extraction. Specifically, \textit{KIEMP} estimates the importance of phrase with three modules: a chunking module to measure its syntactic accuracy, a ranking module to check its information saliency, and a matching module to judge the concept (i.e., topic) consistency between phrase and the whole document. These three modules are seamlessly jointed together via an end-to-end multi-task learning model, which is helpful for three parts to enhance each other and balance the effects of three perspectives. Experimental results on six benchmark datasets show that \textit{KIEMP} outperforms the existing state-of-the-art keyphrase extraction approaches in most cases.

cs.CL↗

arXiv2023

The environmental dependence of Spitzer dusty Supernovae

Thanks to the mid-infrared capability offered by Spitzer, systematic searches of dust in SNe have been carried out over the past decade. Studies have revealed the presence of a substantial amount of dust over a broad range of SN subtypes. How normal SNe present mid-IR excess at later time and turn out to be dusty SNe can be affected by several factors, such as mass-loss history and envelope structure of progenitors and their explosion environment. All these can be combined and related to their environmental properties. A systematic analysis of SNe that exploded under a dusty environment could be of critical importance to measure the properties of the dust-veiled exploding stars, and whether such an intense dust production process is associated with the local environment. In this work, we firstly use the IFS data to study the environmental properties of dusty SNe compared to those of normal ones, and analyze correlations between the environmental properties and their dust parameters. We find that dusty SNe have a larger proportion located at higher SFR regions compared to the normal types. The occurrence of dusty SNe is less dependent on metallicity, with the oxygen abundance spanning from subsolar to oversolar metallicity. We also find the host extinction of dusty SNe scatters a lot, with about 40% of dusty SN located at extremely low extinction environments, and another 30% of them with considerably high host extinction of E(B-V)>0.6 mag.

astro-ph.HE↗

arXiv2023

Serendipitous detection of the dusty Type IIL SN 1980K with JWST/MIRI

We present mid-infrared (mid-IR) imaging of the Type IIL supernova (SN) 1980K with the James Webb Space Telescope (JWST) more than 40 yr post-explosion. SN 1980K, located in the nearby ($D\approx7$ Mpc) "SN factory" galaxy NGC 6946, was serendipitously captured in JWST/MIRI images taken of the field of SN 2004et in the same galaxy. SN 1980K serves as a promising candidate for studying the transitional phase between young SNe and older SN remnants and also provides a great opportunity to investigate its the close environment. SN 1980K can be identified as a clear and bright point source in all eight MIRI filters from F560W up to F2550W. We fit analytical dust models to the mid-IR spectral energy distribution that reveal a large amount ($M_d \approx 0.002 {M}_{\odot}$) of Si-dominated dust at $T_{dust}\approx 150$ K (accompanied by a hotter dust/gas component), and also computed numerical SED dust models. Radiative transfer modeling of a late-time optical spectrum obtained recently with Keck discloses that an even larger ($\sim 0.24-0.58~{M}_{\odot}$) amount of dust is needed in order for selective extinction to explain the asymmetric line profile shapes observed in SN 1980K. As a conclusion, with JWST, we may see i) pre-existing circumstellar dust heated collisionally (or, partly radiatively), analogous to the equatorial ring of SN 1987A, or ii) the mid-IR component of the presumed newly-formed dust, accompanied by much more colder dust present in the ejecta (as suggested by the late-time the optical spectra).

astro-ph.HE↗

arXiv2023

A Superluminous Supernova Lightened by Collisions with Pulsational Pair-instability Shells

Superluminous supernovae are among the most energetic stellar explosions in the Universe, but their energy sources remain an open question. Here we present long-term observations of one of the closest examples of the hydrogen-poor subclass (SLSNe-I), SN~2017egm, revealing the most complicated known luminosity evolution of SLSNe-I. Three distinct post-peak bumps were recorded in its light curve collected at about $100$--350\,days after maximum brightness, challenging current popular power models such as magnetar, fallback accretion, and interaction between ejecta and a circumstellar shell. However, the complex light curve can be well modelled by successive interactions with multiple circumstellar shells with a total mass of about $6.8$--7.7\,M$_\odot$. In this scenario, large energy deposition from interaction-induced reverse shocks results in ionization of neutral oxygen in the supernova ejecta and hence a much lower nebular-phase line ratio of [O\,\textsc{i}] $\lambda6300$/([Ca\,\textsc{ii}] + [O\,\textsc{ii}]) $\lambda7300$ ($\sim 0.2$) compared with that derived for other superluminous and normal stripped-envelope SNe. The pre-existing multiple shells indicate that the progenitor of SN~2017egm experienced pulsational mass ejections triggered by pair instability within 2 years before explosion, in robust agreement with theoretical predictions for a pre-pulsation helium-core mass of 48--51\,M$_{\odot}$. Finally, this work shows that the final explosion product may be a black hole with about 40\,M$_{\odot}$, and has significant implication for the formation of such heavy black holes that have been recently observed by LIGO-Virgo gravitational wave detectors.

astro-ph.HE↗

arXiv2023

Noisy recovery from random linear observations: Sharp minimax rates under elliptical constraints

Estimation problems with constrained parameter spaces arise in various settings. In many of these problems, the observations available to the statistician can be modelled as arising from the noisy realization of the image of a random linear operator; an important special case is random design regression. We derive sharp rates of estimation for arbitrary compact elliptical parameter sets and demonstrate how they depend on the distribution of the random linear operator. Our main result is a functional that characterizes the minimax rate of estimation in terms of the noise level, the law of the random operator, and elliptical norms that define the error metric and the parameter space. This nonasymptotic result is sharp up to an explicit universal constant, and it becomes asymptotically exact as the radius of the parameter space is allowed to grow. We demonstrate the generality of the result by applying it to both parametric and nonparametric regression problems, including those involving distribution shift or dependent covariates.

math.ST↗

arXiv2023

Linear Convergence of Natural Policy Gradient Methods with Log-Linear Policies

We consider infinite-horizon discounted Markov decision processes and study the convergence rates of the natural policy gradient (NPG) and the Q-NPG methods with the log-linear policy class. Using the compatible function approximation framework, both methods with log-linear policies can be written as inexact versions of the policy mirror descent (PMD) method. We show that both methods attain linear convergence rates and $\tilde{\mathcal{O}}(1/ε^2)$ sample complexities using a simple, non-adaptive geometrically increasing step size, without resorting to entropy or other strongly convex regularization. Lastly, as a byproduct, we obtain sublinear convergence rates for both methods with arbitrary constant step size.

cs.LG↗