arXiv subjects

Cheng Cheng

Publications and source records attributed to Cheng Cheng.

At least 91 records · Page 5Linked to original sources

arXiv2025

CogMath: Assessing LLMs' Authentic Mathematical Ability from a Human Cognitive Perspective

Although large language models (LLMs) show promise in solving complex mathematical tasks, existing evaluation paradigms rely solely on a coarse measure of overall answer accuracy, which are insufficient for assessing their authentic capabilities. In this paper, we propose \textbf{CogMath}, which comprehensively assesses LLMs' mathematical abilities through the lens of human cognition. Specifically, inspired by psychological theories, CogMath formalizes human reasoning process into 3 stages: \emph{problem comprehension}, \emph{problem solving}, and \emph{solution summarization}. Within these stages, we investigate perspectives such as numerical calculation, knowledge, and counterfactuals, and design a total of 9 fine-grained evaluation dimensions. In each dimension, we develop an ``\emph{Inquiry}-\emph{Judge}-\emph{Reference}'' multi-agent system to generate inquiries that assess LLMs' mastery from this dimension. An LLM is considered to truly master a problem only when excelling in all inquiries from the 9 dimensions. By applying CogMath on three benchmarks, we reveal that the mathematical capabilities of 7 mainstream LLMs are overestimated by 30\%-40\%. Moreover, we locate their strengths and weaknesses across specific stages/dimensions, offering in-depth insights to further enhance their reasoning abilities.

cs.AI↗

arXiv2025

Towards Foundation Model on Temporal Knowledge Graph Reasoning

Temporal Knowledge Graphs (TKGs) store temporal facts with quadruple formats (s, p, o, t). Existing Temporal Knowledge Graph Embedding (TKGE) models perform link prediction tasks in transductive or semi-inductive settings, which means the entities, relations, and temporal information in the test graph are fully or partially observed during training. Such reliance on seen elements during inference limits the models' ability to transfer to new domains and generalize to real-world scenarios. A central limitation is the difficulty in learning representations for entities, relations, and timestamps that are transferable and not tied to dataset-specific vocabularies. To overcome these limitations, we introduce the first fully-inductive approach to temporal knowledge graph link prediction. Our model employs sinusoidal positional encodings to capture fine-grained temporal patterns and generates adaptive entity and relation representations using message passing conditioned on both local and global temporal contexts. Our model design is agnostic to temporal granularity and time span, effectively addressing temporal discrepancies across TKGs and facilitating time-aware structural information transfer. As a pretrained, scalable, and transferable model, POSTRA demonstrates strong zero-shot performance on unseen temporal knowledge graphs, effectively generalizing to novel entities, relations, and timestamps. Extensive theoretical analysis and empirical results show that a single pretrained model can improve zero-shot performance on various inductive temporal reasoning scenarios, marking a significant step toward a foundation model for temporal KGs.

cs.AI↗

arXiv2025

HaploOmni: Unified Single Transformer for Multimodal Video Understanding and Generation

With the advancement of language models, unified multimodal understanding and generation have made significant strides, with model architectures evolving from separated components to unified single-model frameworks. This paper explores an efficient training paradigm to build a single transformer for unified multimodal understanding and generation. Specifically, we propose a multimodal warmup strategy utilizing prior knowledge to extend capabilities. To address cross-modal compatibility challenges, we introduce feature pre-scaling and multimodal AdaLN techniques. Integrating the proposed technologies, we present the HaploOmni, a new single multimodal transformer. With limited training costs, HaploOmni achieves competitive performance across multiple image and video understanding and generation benchmarks over advanced unified models. All codes will be made public at https://github.com/Tencent/HaploVLM.

cs.CV↗

arXiv2025

From Objectives to Questions: A Planning-based Framework for Educational Mathematical Question Generation

Automatically generating high-quality mathematical problems that align with educational objectives is a crucial task in NLP-based educational technology. Traditional generation methods focus primarily on textual quality, but they often overlook educational objectives. Moreover, these methods address only single-dimensional, simple question generation, failing to meet complex, multifaceted educational requirements. To address these challenges, we constructed and annotated EduMath, a dataset of 16k mathematical questions with multi-dimensional educational objectives. Based on this dataset, we developed EQGEVAL, which incorporates three evaluation dimensions and is designed to assess the ability of models to generate educational questions. Drawing inspiration from teachers' problem design processes, we propose the Educational Question Planning with self-Reflection (EQPR) method for educational mathematical question generation, following a "plan-evaluate-optimize" approach. Specifically, by combining planning algorithm based on Monte Carlo Tree Search with the generative capabilities of Large Language Models, we continuously optimize questions through iterative feedback. This self-optimization mechanism ensures that the generated questions both fit the educational context and strategically achieve specific basic educational objectives. Through extensive experiments based on EQGEVAL, we have demonstrated that EQPR achieves significant improvements in generating questions that meet multi-dimensional educational objectives.

cs.CL↗

arXiv2025

Skywork Open Reasoner 1 Technical Report

The success of DeepSeek-R1 underscores the significant role of reinforcement learning (RL) in enhancing the reasoning capabilities of large language models (LLMs). In this work, we present Skywork-OR1, an effective and scalable RL implementation for long Chain-of-Thought (CoT) models. Building on the DeepSeek-R1-Distill model series, our RL approach achieves notable performance gains, increasing average accuracy across AIME24, AIME25, and LiveCodeBench from 57.8% to 72.8% (+15.0%) for the 32B model and from 43.6% to 57.5% (+13.9%) for the 7B model. Our Skywork-OR1-32B model surpasses both DeepSeek-R1 and Qwen3-32B on the AIME24 and AIME25 benchmarks, while achieving comparable results on LiveCodeBench. The Skywork-OR1-7B and Skywork-OR1-Math-7B models demonstrate competitive reasoning capabilities among models of similar size. We perform comprehensive ablation studies on the core components of our training pipeline to validate their effectiveness. Additionally, we thoroughly investigate the phenomenon of entropy collapse, identify key factors affecting entropy dynamics, and demonstrate that mitigating premature entropy collapse is critical for improved test performance. To support community research, we fully open-source our model weights, training code, and training datasets.

cs.LG↗

arXiv2025

AD-AGENT: A Multi-agent Framework for End-to-end Anomaly Detection

Anomaly detection (AD) is essential in areas such as fraud detection, network monitoring, and scientific research. However, the diversity of data modalities and the increasing number of specialized AD libraries pose challenges for non-expert users who lack in-depth library-specific knowledge and advanced programming skills. To tackle this, we present AD-AGENT, an LLM-driven multi-agent framework that turns natural-language instructions into fully executable AD pipelines. AD-AGENT coordinates specialized agents for intent parsing, data preparation, library and model selection, documentation mining, and iterative code generation and debugging. Using a shared short-term workspace and a long-term cache, the agents integrate popular AD libraries like PyOD, PyGOD, and TSLib into a unified workflow. Experiments demonstrate that AD-AGENT produces reliable scripts and recommends competitive models across libraries. The system is open-sourced to support further research and practical applications in AD.

cs.CL↗

arXiv2025

FAST Reveals the Extended HI Halo and Accretion Signatures of NGC2683

We present the results of our recent HI observations conducted on the edge-on galaxy NGC2683 using the Five-hundred-meter Aperture Spherical radio Telescope (FAST). In comparison to previous observations made by the VLA, FAST has detected a more extensive distribution of HI. Particularly noteworthy is that the detections made by FAST extend approximately four times farther than those of the VLA in the vertical direction from the galactic plane. The total HI flux measured for NGC2683 amounts to $F_{\rm HI} = 112.1\,\rm{Jy\,km\,s^{-1}}$ (equivalent to a total HI mass of $M_{\rm HI} = 2.32 \times 10^9\,{\rm M_\odot}$), which is slightly higher than that detected by VLA. FAST has also identified three dwarf galaxies in close proximity to NGC2683, namely KK69, NGC2683dw1 (hereafter dw1), and NGC2683dw3$?$ (hereafter dw3$?$). dw3$?$ is situated within the extended HI distribution of NGC2683 in projection and lies near the tail of KK69 extending towards NGC2683. These observations suggest that dw3$?$ is likely a result of the accretion process from NGC2683 to KK69. Furthermore, FAST has detected three high-velocity clouds (HVCs), with complex B potentially undergoing accretion with NGC2683. Based on the model from Vollmer et al. 2016 and incorporating the HI halo component, we found that the model with the added HI halo aligns more closely with our FAST observations in NGC2683. The estimated mass of this HI halo is $3 \times 10^8\,{\rm M_\odot}$, constituting approximately 13% of the total HI mass of the galaxy. We suggest that the origination of this HI halo is more likely attributed to external gas accretion.

astro-ph.GA↗

arXiv2025

Iterative Polynomial Approximation Algorithms for Inverse Graph Filters

Chebyshev interpolation polynomials exhibit the exponential approximation property to analytic functions on a cube. Based on the Chebyshev interpolation polynomial approximation, we propose iterative polynomial approximation algorithms to implement the inverse filter with a polynomial graph filter of commutative graph shifts in a distributed manner. The proposed algorithms exhibit exponential convergence properties, and they can be implemented on distributed networks in which agents are equipped with a data processing subsystem for limited data storage and computation power, and with a one-hop communication subsystem for direct data exchange only with their adjacent agents. Our simulations show that the proposed polynomial approximation algorithms may converge faster than the Chebyshev polynomial approximation algorithm and the conventional gradient descent algorithm do.

eess.SP↗

arXiv2025

Medium-band Astrophysics with the Grism of NIRCam In Frontier fields (MAGNIF): Spectroscopic Census of H$α$ Luminosity Functions and Cosmic Star Formation at $z\sim 4.5$ and 6.3

We measure H$α$ luminosity functions (LFs) at redshifts $z \sim 4.5$ and 6.3 using the JWST MAGNIF (Medium-band Astrophysics with the Grism of NIRCam In Frontier fields) survey. MAGNIF obtained NIRCam grism spectra with the F360M and F480M filters in four Frontier Fields. We identify 248 H$α$ emitters based on the grism spectra and photometric redshifts from combined HST and JWST imaging data. The numbers of the H$α$ emitters show a large field-to-field variation, highlighting the necessity of multiple fields to mitigate cosmic variance. We calculate both observed and dust-corrected H$α$ LFs in the two redshift bins. Thanks to the gravitational lensing, the measured H$α$ LFs span three orders of magnitude in luminosity, and the faint-end luminosity reaches $L_{\mathrm{H}α} \sim 10^{40.3} \mathrm{erg} \mathrm{s}^{-1}$ at $z \sim 4.5$ and $10^{41.5} \mathrm{erg} \mathrm{s}^{-1}$ at $z \sim 6.3$, corresponding to star-formation rates (SFRs) of $\sim$ 0.1 and 1.7 $\mathrm{M}_\odot \mathrm{yr}^{-1}$. We conclude no or weak redshift evolution of the faint-end slope of H$α$ LF across $z\simeq0.4-6.3$, and the comparison with the faint-end slopes of UV LF indicates stochastic star formation history among low-mass H$α$ emitters. The derived cosmic SFR densities are $0.058^{+0.008}_{-0.006}\ \ M_\odot\ \mathrm{yr}^{-1}\ \mathrm{Mpc}^{-3}$ at $z \sim 4.5$ and $0.025^{+0.009}_{-0.007}\ \ M_\odot\ \mathrm{yr}^{-1}\ \mathrm{Mpc}^{-3}$ at $z \sim 6.3$. These are approximately 2.2 times higher than previous estimates based on dust-corrected UV LFs, but consistent with recent measurements from infrared surveys. We discuss uncertainties in the H$α$ LF measurements, including those propagate from the lens models, cosmic variance, and AGN contribution.

astro-ph.GA↗

arXiv2025

Improving vision-language alignment with graph spiking hybrid Networks

To bridge the semantic gap between vision and language (VL), it is necessary to develop a good alignment strategy, which includes handling semantic diversity, abstract representation of visual information, and generalization ability of models. Recent works use detector-based bounding boxes or patches with regular partitions to represent visual semantics. While current paradigms have made strides, they are still insufficient for fully capturing the nuanced contextual relations among various objects. This paper proposes a comprehensive visual semantic representation module, necessitating the utilization of panoptic segmentation to generate coherent fine-grained semantic features. Furthermore, we propose a novel Graph Spiking Hybrid Network (GSHN) that integrates the complementary advantages of Spiking Neural Networks (SNNs) and Graph Attention Networks (GATs) to encode visual semantic information. Intriguingly, the model not only encodes the discrete and continuous latent variables of instances but also adeptly captures both local and global contextual features, thereby significantly enhancing the richness and diversity of semantic representations. Leveraging the spatiotemporal properties inherent in SNNs, we employ contrastive learning (CL) to enhance the similarity-based representation of embeddings. This strategy alleviates the computational overhead of the model and enriches meaningful visual representations by constructing positive and negative sample pairs. We design an innovative pre-training method, Spiked Text Learning (STL), which uses text features to improve the encoding ability of discrete semantics. Experiments show that the proposed GSHN exhibits promising results on multiple VL downstream tasks.

cs.CV↗

arXiv2025

EPOCHS IV: SED Modelling Assumptions and their impact on the Stellar Mass Function at 6.5 < z < 13.5 using PEARLS and public JWST observations

We utilize deep JWST NIRCam observations for the first direct constraints on the Galaxy Stellar Mass Function (GSMF) at $z>10$. Our EPOCHS v1 sample includes 1120 galaxy candidates at $6.5<z<13.5$ taken from a consistent reduction and analysis of publicly available deep JWST NIRCam data covering the PEARLS, CEERS, GLASS, JADES GOOD-S, NGDEEP, and SMACS0723 surveys, totalling 187 arcmin$^2$. We investigate the impact of SED fitting methods, assumed star formation histories (SFH), dust laws, and priors on galaxy masses and the resultant GSMF. Whilst our fiducial GSMF agrees with the literature at $z<13.5$, we find that the assumed SFH model has a large impact on the GSMF and stellar mass density (SMD), finding a 0.75~dex increase in the SMD at $z=10.5$ between a flexible non-parametric and standard parametric SFH. Overall, we find a flatter SMD evolution at $z \geq 9$ than some studies predict, suggesting a rapid buildup of stellar mass in the early Universe. We find no incompatibility between our results and those of standard cosmological models, as suggested previously, although the most massive galaxies may require a high star formation efficiency. We find that the "Little Red Dot" galaxies dominate the $z=7$ GSMF at high-masses, necessitating a better understanding of the relative contributions of AGN and stellar emission. We show that assuming a theoretically motivated top-heavy IMF reduces stellar mass by 0.5~dex without affecting fit quality, but our results remain consistent with existing cosmological models with a standard IMF.

astro-ph.GA↗

arXiv2025

Starburst Galaxies in Their Last Billion Years: An H$δ$ Absorption Line Selected Sample

In this paper, we focus on the study of starburst galaxies in their final billion years. Our galaxy selection is based solely on the presence of the H$δ$ absorption line, which permits tracing the later evolution of starburst galaxies, coinciding with the emergence of A-type stars in these galaxies. We propose a novel method that utilizes star formation rate and UVJ colors to classify galaxies in the sample, and use the spectral features to mark their evolution stages. Our in-depth analysis of the MgII line indicates the substantial increasing of F- and G-type stars when a galaxy evolves from star forming to quiescent phase. Furthermore, we identify AGNs in this sample to explore their roles in the later stage of galaxy star formation history.

astro-ph.GA↗

arXiv2025

The Gas-to-Dust Ratio Investigation in the Massive Star-Forming region M17

M17 is a well-known massive star-forming region, and its Gas-to-Dust Ratio (GDR) may vary significantly compared to the other areas. The mass of gas can be traced by the ${\rm CO}$ emission observed in the \emph{Milky Way Imaging Scroll Painting (MWISP) project}. The dust mass can be traced by analyzing the interstellar extinction magnitude obtained from the \emph{United Kingdom Infrared Telescope (UKIRT)}. We computed the ratio ${W({\rm CO})/A_V}$: for ${A_V \le }$ 10 mag, ${{ W(^{12}{\rm CO})/ A_V}= (6.27 \pm 0.19)}$ ${\mathrm{{K \cdot km/s} \cdot mag^{-1}}}$ and ${{ W(^{13}{\rm CO})/ A_V} = (0.75 \pm 0.72)}$ ${ \mathrm{{K \cdot km/s} \cdot mag^{-1}}}$; whereas for ${{A_V} \ge 10}$ mag, ${{ W(^{12}{\rm CO})/ A_V} = (15.8 \pm 0.06) }$ ${\mathrm{{K \cdot km/s} \cdot mag^{-1}}}$ and ${{ W(^{13}{\rm CO})/ A_V} = (3.11 \pm 0.25)}$ ${ \mathrm{{K \cdot km/s} \cdot mag^{-1}}}$. Then, we converted the ${W({\rm CO})/A_V}$ into ${N(\rm H)/A_V}$. Using the WD01 model, we derived the GDR: for ${A_V \le }$ 10 mag, the GDRs were ${118 \pm 9}$ for ${^{12}{\rm CO}}$ and ${83 \pm 62}$ for ${^{13}{\rm CO}}$, comparable to those of the Milky Way; however, for ${A_V \ge }$ 10 mag, the GDRs increased significantly to ${296 \pm 3}$ for ${^{12}{\rm CO}}$ and ${387 \pm 40}$ for ${^{13}{\rm CO}}$, approximately three times higher than those of the Milky Way. In the discussion, we compared the results of this work with previous studies and provided a detailed discussion of the influence of massive stars and other factors on GDR.

astro-ph.GA↗

arXiv2025

Comparison of Global HI and H$α$ Line Profiles in MaNGA Galaxy Pairs with FAST

We present case studies comparing the global HI and H$α$ emission line profiles of six galaxy pairs. The six pairs are selected to have different nuclear activities, with two hosting an active galactic nucleus, and in different merging stages (two of each from pre-merging, merging, and post-merger stages). We observe their global HI spectra with the Five-hundred-meter Aperture Spherical radio Telescope (FAST), achieving a noise level of about 0.5 mJy. Five out of the six pair systems have secure detections of HI emissions (signal-to-noise ratio > 10). The HI fraction and star formation efficiency of the six pairs do not deviate from isolated galaxies. For the HI line profiles, common unique asymmetry is observed, indicating disturbances on the atomic gas from the galaxy interaction. The global H$α$ spectra of the merger systems are constructed from the optical integral field spectroscopic observations, by integrating the flux in corresponding line-of-sight velocity bins. The H$α$ spectra tend to show multiple components in the pre-merger phase, and single component line profiles in the post-merger systems, while all HI spectra show single component line profiles regardless of merger stages. The HI and H$α$ spectra show offsets in the central velocities, which appear to decrease from >100 km/s in the pre-merger pair to <10 km/s in post-merger pairs. This trend is consistent with the scenario that, despite the significantly different distribution and kinematics of the atomic and ionized gases, the merging process may contribute to the mixing and eventually align various gas contents.

astro-ph.GA↗

arXiv2024

Automatic Spectral Calibration of Hyperspectral Images:Method, Dataset and Benchmark

Hyperspectral image (HSI) densely samples the world in both the space and frequency domain and therefore is more distinctive than RGB images. Usually, HSI needs to be calibrated to minimize the impact of various illumination conditions. The traditional way to calibrate HSI utilizes a physical reference, which involves manual operations, occlusions, and/or limits camera mobility. These limitations inspire this paper to automatically calibrate HSIs using a learning-based method. Towards this goal, a large-scale HSI calibration dataset is created, which has 765 high-quality HSI pairs covering diversified natural scenes and illuminations. The dataset is further expanded to 7650 pairs by combining with 10 different physically measured illuminations. A spectral illumination transformer (SIT) together with an illumination attention module is proposed. Extensive benchmarks demonstrate the SoTA performance of the proposed SIT. The benchmarks also indicate that low-light conditions are more challenging than normal conditions. The dataset and codes are available online:https://github.com/duranze/Automatic-spectral-calibration-of-HSI

cs.CV↗

arXiv2024

Ultra Diffuse Dwarf Galaxies Hosting Pseudo-bulges

By analyzing data from DESI Legacy Imaging Survey of the dwarf galaxies in the Arecibo Legacy Fast Alfa Survey, we have identified five ultra-diffuse galaxies (UDGs) featuring central pseudo-bulges. These UDGs display blue pseudo-bulges with Sérsic indices $n<2.5$ and effective radii spanning 300-700 pc, along with bluer thin stellar disks exhibiting low surface brightness and expansive effective radii that align with the UDG definition. The rotation velocities of these UDGs, determined using HI line widths and optical inclinations, exceed those of most dwarf galaxies of similar mass, suggesting the high halo spins or substantial dark matter halos. We propose that these UDGs likely formed through mergers of dwarf galaxies lacking old stars in their progenitors, resulting in the development of central bulge-like structures during starbursts triggered by the mergers, while also enhancing their halo spin. Subsequent gas accretion facilitated the formation of extended stellar disks. It is also worth noting the possibility that these UDGs could alternatively represent ``failed $L^{\star}$ galaxies'' with massive dark matter halos but reduced star formation efficiencies. If future high-resolution HI observations confirm the presence of massive halos around these UDGs, they may have formed due to intense AGN feedback in the early universe, and may be the descendants of ``little red dots'' observed by the James Webb Space Telescope, which are characterized by heightened central black hole masses and intensified accretion and feedback processes in the early universe.

astro-ph.GA↗

arXiv2024

A Short-lived Rejuvenation during the Decades-long Changing-look Transition in the Nucleus of Mrk 1018

Changing-look active galactic nuclei (CL-AGNs), characterized by emerging or disappearing of broad lines accompanied with extreme continuum flux variability, have drawn much attention for their potential of revealing physical processes underlying AGN evolution. We perform seven-season spectroscopic monitoring on Mrk~1018, one of the earliest identified CL-AGN. Around 2020, we detect a full-cycle changing-look transition of Mrk~1018 within one year, associated with a nucleus outburst, which likely arise from the disk instability in the transition region between the outer standard rotation-dominated disk and inner advection-dominated accretion flow. Over the past forty-five years, the accretion rate of Mrk~1018 changed 1000 times and the maximum Eddington ratio reached 0.02. By investigating the relation between broad-line properties and Eddington ratio ($L_{\rm bol}/L_{\rm Edd}$), we find strong evidence that the full-cycle type transition is regulated by accretion. There exists a turnover point in the Balmer decrement, which is observed for the first time. The broad Balmer lines change from a single peak in Type 1.0-1.2 to double peaks in Type 1.5-1.8 and the double-peak separation decreases with increasing accretion rate. We also find that the full width at half maximum (FWHM) of the broad Balmer lines obeys FWHM$\propto (L_{\rm bol}/L_{\rm Edd})^{-0.27}$, as expected for a virialized BLR. The velocity dispersion $σ_{\rm line}$ follows a similar trend in Type 1.5-1.8, but displays a sharp increases in Type 1.0-1.2, resulting in a dramatic drop of FWHM/$σ_{\rm line}$. These findings suggest that a virialized BLR together with accretion-dependent turbulent motions might be responsible for the diversity of BLR phenomena across AGN population.

astro-ph.GA↗

arXiv2024

PLReMix: Combating Noisy Labels with Pseudo-Label Relaxed Contrastive Representation Learning

Recently, the usage of Contrastive Representation Learning (CRL) as a pre-training technique improves the performance of learning with noisy labels (LNL) methods. However, instead of pre-training, when trivially combining CRL loss with LNL methods as an end-to-end framework, the empirical experiments show severe degeneration of the performance. We verify through experiments that this issue is caused by optimization conflicts of losses and propose an end-to-end \textbf{PLReMix} framework by introducing a Pseudo-Label Relaxed (PLR) contrastive loss. This PLR loss constructs a reliable negative set of each sample by filtering out its inappropriate negative pairs, alleviating the loss conflicts by trivially combining these losses. The proposed PLR loss is pluggable and we have integrated it into other LNL methods, observing their improved performance. Furthermore, a two-dimensional Gaussian Mixture Model is adopted to distinguish clean and noisy samples by leveraging semantic information and model outputs simultaneously. Experiments on multiple benchmark datasets demonstrate the effectiveness of the proposed method. Code is available at \url{https://github.com/lxysl/PLReMix}.

cs.CV↗