Searcharxiv⌕ Search

arXiv subjects

Li Xiao

Publications and source records attributed to Li Xiao.

At least 37 records · Page 2Linked to original sources
arXiv2025

UniDCF: A Foundation Model for Comprehensive Dentocraniofacial Hard Tissue Reconstruction

Dentocraniofacial hard tissue defects profoundly affect patients' physiological functions, facial aesthetics, and psychological well-being, posing significant challenges for precise reconstruction. Current deep learning models are limited to single-tissue scenarios and modality-specific imaging inputs, resulting in poor generalizability and trade-offs between anatomical fidelity, computational efficiency, and cross-tissue adaptability. Here we introduce UniDCF, a unified framework capable of reconstructing multiple dentocraniofacial hard tissues through multimodal fusion encoding of point clouds and multi-view images. By leveraging the complementary strengths of each modality and incorporating a score-based denoising module to refine surface smoothness, UniDCF overcomes the limitations of prior single-modality approaches. We curated the largest multimodal dataset, comprising intraoral scans, CBCT, and CT from 6,609 patients, resulting in 54,555 annotated instances. Evaluations demonstrate that UniDCF outperforms existing state-of-the-art methods in terms of geometric precision, structural completeness, and spatial accuracy. Clinical simulations indicate UniDCF reduces reconstruction design time by 99% and achieves clinician-rated acceptability exceeding 94%. Overall, UniDCF enables rapid, automated, and high-fidelity reconstruction, supporting personalized and precise restorative treatments, streamlining clinical workflows, and enhancing patient outcomes.

cs.CV↗
arXiv2025

FreeCodec: A disentangled neural speech codec with fewer tokens

Neural speech codecs have gained great attention for their outstanding reconstruction with discrete token representations. It is a crucial component in generative tasks such as speech coding and large language models (LLM). However, most works based on residual vector quantization perform worse with fewer tokens due to low coding efficiency for modeling complex coupled information. In this paper, we propose a neural speech codec named FreeCodec which employs a more effective encoding framework by decomposing intrinsic properties of speech into different components: 1) a global vector is extracted as the timbre information, 2) a prosody encoder with a long stride level is used to model the prosody information, 3) the content information is from a content encoder. Using different training strategies, FreeCodec achieves state-of-the-art performance in reconstruction and disentanglement scenarios. Results from subjective and objective experiments demonstrate that our framework outperforms existing methods.

cs.SD↗
arXiv2025

Sparse Phase Retrieval with Redundant Dictionary via $\ell_q (0<q\le 1)$-Analysis Model

Sparse phase retrieval with redundant dictionary is to reconstruct the signals of interest that are (nearly) sparse in a redundant dictionary or frame from the phaseless measurements via the optimization models. Gao [7] presented conditions on the measurement matrix, called null space property (NSP) and strong dictionary restricted isometry property (S-DRIP), for exact and stable recovery of dictionary-$k$-sparse signals via the $\ell_1$-analysis model for sparse phase retrieval with redundant dictionary, respectively, where, in particularly, the S-DRIP of order $tk$ with $t>1$ was derived. In this paper, motivated by many advantages of the $\ell_q$ minimization with $0<q\leq1$, e.g., reduction of the number of measurements required, we generalize these two conditions to the $\ell_q$-analysis model. Specifically, we first present two NSP variants for exact recovery of dictionary-$k$-sparse signals via the $\ell_q$-analysis model in the noiseless scenario. Moreover, we investigate the S-DRIP of order $tk$ with $0<t<\frac{4}{3}$ for stable recovery of dictionary-$k$-sparse signals via the $\ell_q$-analysis model in the noisy scenario, which will complement the existing result of the S-DRIP of order $tk$ with $t\geq2$ obtained in [4].

cs.IT↗
arXiv2025

A candidate for True Type-2 AGN without hidden central BLRs Identified by central Tidal Disruption Event

In this manuscript, through applications of TDE (tidal disruption event) expected variability properties, a potential candidate for True type-2 AGN without hidden central broad line regions (=TT2AGN) is reported in the SDSS J233454.07+145712.9 (=SDSS J2334). Through analyzing the 20-years optical light curves of SDSS J2334 from different Sky Survey projects, a TDE is preferred with a $4.7{\rm M_\odot}$ main-sequence star tidally disrupted by the central BH with mass $11.7\times 10^6{\rm M_\odot}$, indicating that central region within distance about 20 light-days to central BH in SDSS J2334 is directly in the line-of-sight. Moreover, AGN activities in SDSS J2334 can be confirmed through applications of BPT diagrams. Meanwhile, comparing virial BH mass determined through assumed broad Balmer emission components and M-sigma expected BH mass by well measured stellar velocity dispersion through stellar absorption features, optical broad emission lines in SDSS J2334 are disfavored with confidence level higher than 6$σ$. Therefore, combining the unique properties of the TDE and the spectroscopic results with only narrow emission lines, SDSS J2334 can be well identified as a potential candidate for a TT2AGN. The results indicate the to detect TDE expected flares in normal Type-2 AGN classified by spectroscopic results should be a new practicable method for identifying

astro-ph.GA↗
arXiv2025

Human-Aligned Bench: Fine-Grained Assessment of Reasoning Ability in MLLMs vs. Humans

The goal of achieving Artificial General Intelligence (AGI) is to imitate humans and surpass them. Models such as OpenAI's o1, o3, and DeepSeek's R1 have demonstrated that large language models (LLMs) with human-like reasoning capabilities exhibit exceptional performance and are being gradually integrated into multimodal large language models (MLLMs). However, whether these models possess capabilities comparable to humans in handling reasoning tasks remains unclear at present. In this paper, we propose Human-Aligned Bench, a benchmark for fine-grained alignment of multimodal reasoning with human performance. Specifically, we collected 9,794 multimodal questions that solely rely on contextual reasoning, including bilingual (Chinese and English) multimodal questions and pure text-based questions, encompassing four question types: visual reasoning, definition judgment, analogical reasoning, and logical judgment. More importantly, each question is accompanied by human success rates and options that humans are prone to choosing incorrectly. Extensive experiments on the Human-Aligned Bench reveal notable differences between the performance of current MLLMs in multimodal reasoning and human performance. The findings on our benchmark provide insights into the development of the next-generation models.

cs.CV↗
arXiv2025

A Scalable Crawling Algorithm Utilizing Noisy Change-Indicating Signals

Web refresh crawling is the problem of keeping a cache of web pages fresh, that is, having the most recent copy available when a page is requested, given a limited bandwidth available to the crawler. Under the assumption that the change and request events, resp., to each web page follow independent Poisson processes, the optimal scheduling policy was derived by Azar et al. 2018. In this paper, we study an extension of this problem where side information indicating content changes, such as various types of web pings, for example, signals from sitemaps, content delivery networks, etc., is available. Incorporating such side information into the crawling policy is challenging, because (i) the signals can be noisy with false positive events and with missing change events; and (ii) the crawler should achieve a fair performance over web pages regardless of the quality of the side information, which might differ from web page to web page. We propose a scalable crawling algorithm which (i) uses the noisy side information in an optimal way under mild assumptions; (ii) can be deployed without heavy centralized computation; (iii) is able to crawl web pages at a constant total rate without spikes in the total bandwidth usage over any time interval, and automatically adapt to the new optimal solution when the total bandwidth changes without centralized computation. Experiments clearly demonstrate the versatility of our approach.

stat.ML↗
arXiv2025

Magnetic field properties of the SNR HB 9

We aim to study the polarization and magnetic field properties of the SNR HB 9 using new 21-cm continuum cube data from the Five-hundred-meter Aperture Spherical radio telescope (FAST). We computed the Faraday depth at 21 cm, and re-analyzed the rotation measures (RMs) of HB 9 using in addition Effelsberg 2695-MHz and Urumqi 4800-MHz polarization data. FAST total-intensity images of two subbands are decomposed into components of multiple angular scales to check spectral-index variation via temperature versus temperature plots (TT-plots). The filamentary emission has a spectral index ($S\simν^α$) of $α=-$0.52, corresponding to freshly accelerated relativistic electrons. The diffuse emission has a steeper spectrum of $α=-$0.63, corresponding to confined electrons that are no longer accelerated. The FAST detected 1385-MHz polarized emission might come from a thin layer in the outer envelope of the shells, with a Faraday depth of 4-28 rad m$^{-2}$ from the Faraday rotation synthesis result. The RMs derived from the Effelsberg 2695-MHz and Urumqi 4800-MHz polarization data show about 70 rad m$^{-2}$ in the eastern and northern shell, and 124 rad m$^{-2}$ in the inner and southern patches. The regular magnetic field is about 5$-$8 $μ$G over the remnant. The northern shell shows depolarization at 2695 MHz relative to the 4800-MHz polarization data, indicating an additional random magnetic field of 12 $μ$G on the scale of 0.6 pc. The shock wave might have entered the dense gas environment in the northern-shell region and has driven turbulence to cause depolarization at 2695 MHz.

astro-ph.GA↗
arXiv2025

Learning 3D Medical Image Models From Brain Functional Connectivity Network Supervision For Mental Disorder Diagnosis

In MRI-based mental disorder diagnosis, most previous studies focus on functional connectivity network (FCN) derived from functional MRI (fMRI). However, the small size of annotated fMRI datasets restricts its wide application. Meanwhile, structural MRIs (sMRIs), such as 3D T1-weighted (T1w) MRI, which are commonly used and readily accessible in clinical settings, are often overlooked. To integrate the complementary information from both function and structure for improved diagnostic accuracy, we propose CINP (Contrastive Image-Network Pre-training), a framework that employs contrastive learning between sMRI and FCN. During pre-training, we incorporate masked image modeling and network-image matching to enhance visual representation learning and modality alignment. Since the CINP facilitates knowledge transfer from FCN to sMRI, we introduce network prompting. It utilizes only sMRI from suspected patients and a small amount of FCNs from different patient classes for diagnosing mental disorders, which is practical in real-world clinical scenario. The competitive performance on three mental disorder diagnosis tasks demonstrate the effectiveness of the CINP in integrating multimodal MRI information, as well as the potential of incorporating sMRI into clinical diagnosis using network prompting.

cs.CV↗
arXiv2024

Advancing Fine-Grained Visual Understanding with Multi-Scale Alignment in Multi-Modal Models

Multi-modal large language models (MLLMs) have achieved remarkable success in fine-grained visual understanding across a range of tasks. However, they often encounter significant challenges due to inadequate alignment for fine-grained knowledge, which restricts their ability to accurately capture local details and attain a comprehensive global perception. While recent advancements have focused on aligning object expressions with grounding information, they typically lack explicit integration of object images, which contain affluent information beyond mere texts or coordinates. To bridge this gap, we introduce a novel fine-grained visual knowledge alignment method that effectively aligns and integrates multi-scale knowledge of objects, including texts, coordinates, and images. This innovative method is underpinned by our multi-scale fine-grained enhancement data synthesis pipeline, which provides over 300K essential training data to enhance alignment and improve overall performance. Furthermore, we present TinyGroundingGPT, a series of compact models optimized for high-level alignments. With a scale of approximately 3B parameters, TinyGroundingGPT achieves outstanding results in grounding tasks while delivering performance comparable to larger MLLMs in complex visual scenarios.

cs.CV↗
arXiv2024

Service Deployment in the On-Demand Economy: Employees, Contractors, or Both?

The recent advancements in mobile/data technology have fostered a widespread adoption of on-demand or gig service platforms. The increasingly available data and independent contractors have enabled these platforms to design customized services and a cost-efficient workforce to effectively match demand and supply. In practice, a diverse landscape of the workforce has been observed: some rely solely on either employees or contractors, others use a blended workforce with both types of workers. In this paper, we consider a profit-maximizing service provider (SP) that decides to offer a single service or two differentiated services, along with the pricing and staffing of the workforce with employees and/or contractors, to price- and waiting-sensitive customers. Contractors independently determine whether or not to participate in the marketplace based on private reservation rates and per-service wage offered by the SP, while it controls the number of employees who receive per-hour wage. Under a single service, we show that the SP relies on either employees or contractors and identify sufficient and necessary conditions in which one workforce is better than the other. Under the optimal service deployment, we show that the SP offers either a single service relying solely on employees or contractors, or two differentiated services with a hybrid workforce depending on the service value and cost efficiencies of employees and contractors. Our analysis suggests that proliferating services with a blended workforce could improve the SP's profit significantly, and identifies conditions in which this value is significant. Our results provide an in-depth understanding and insightful guidance to on-demand platforms on the design of service differentiation and workforce models.

math.OC↗
arXiv2024

Adapter-dependent Adapter Methylation Assay

Sensitive and reliable methylation assay is important for oncogentic studies and clinical applications. Here, a new methylation assay was developed by the use of adapter-dependent adapter in library preparation. This new assay avoids the use of bisulfite and provides a simple and highly sensitive scanning of methylation spectra of circulating free DNA and genomic DNA.

q-bio.BM↗
arXiv2024

QCRD: Quality-guided Contrastive Rationale Distillation for Large Language Models

The deployment of large language models (LLMs) faces considerable challenges concerning resource constraints and inference efficiency. Recent research has increasingly focused on smaller, task-specific models enhanced by distilling knowledge from LLMs. However, prior studies have often overlooked the diversity and quality of knowledge, especially the untapped potential of negative knowledge. Constructing effective negative knowledge remains severely understudied. In this paper, we introduce a novel framework called quality-guided contrastive rationale distillation aimed at enhancing reasoning capabilities through contrastive knowledge learning. For positive knowledge, we enrich its diversity through temperature sampling and employ self-consistency for further denoising and refinement. For negative knowledge, we propose an innovative self-adversarial approach that generates low-quality rationales by sampling previous iterations of smaller language models, embracing the idea that one can learn from one's own weaknesses. A contrastive loss is developed to distill both positive and negative knowledge into smaller language models, where an online-updating discriminator is integrated to assess qualities of rationales and assign them appropriate weights, optimizing the training process. Through extensive experiments across multiple reasoning tasks, we demonstrate that our method consistently outperforms existing distillation techniques, yielding higher-quality rationales.

cs.CL↗
arXiv2024

SuperCodec: A Neural Speech Codec with Selective Back-Projection Network

Neural speech coding is a rapidly developing topic, where state-of-the-art approaches now exhibit superior compression performance than conventional methods. Despite significant progress, existing methods still have limitations in preserving and reconstructing fine details for optimal reconstruction, especially at low bitrates. In this study, we introduce SuperCodec, a neural speech codec that achieves state-of-the-art performance at low bitrates. It employs a novel back projection method with selective feature fusion for augmented representation. Specifically, we propose to use Selective Up-sampling Back Projection (SUBP) and Selective Down-sampling Back Projection (SDBP) modules to replace the standard up- and down-sampling layers at the encoder and decoder, respectively. Experimental results show that our method outperforms the existing neural speech codecs operating at various bitrates. Specifically, our proposed method can achieve higher quality reconstructed speech at 1 kbps than Lyra V2 at 3.2 kbps and Encodec at 6 kbps.

cs.SD↗
arXiv2024

SimuSOE: A Simulated Snoring Dataset for Obstructive Sleep Apnea-Hypopnea Syndrome Evaluation during Wakefulness

Obstructive Sleep Apnea-Hypopnea Syndrome (OSAHS) is a prevalent chronic breathing disorder caused by upper airway obstruction. Previous studies advanced OSAHS evaluation through machine learning-based systems trained on sleep snoring or speech signal datasets. However, constructing datasets for training a precise and rapid OSAHS evaluation system poses a challenge, since 1) it is time-consuming to collect sleep snores and 2) the speech signal is limited in reflecting upper airway obstruction. In this paper, we propose a new snoring dataset for OSAHS evaluation, named SimuSOE, in which a novel and time-effective snoring collection method is introduced for tackling the above problems. In particular, we adopt simulated snoring which is a type of snore intentionally emitted by patients to replace natural snoring. Experimental results indicate that the simulated snoring signal during wakefulness can serve as an effective feature in OSAHS preliminary screening.

cs.SD↗
arXiv2024

Aligning in a Compact Space: Contrastive Knowledge Distillation between Heterogeneous Architectures

Knowledge distillation is commonly employed to compress neural networks, reducing the inference costs and memory footprint. In the scenario of homogenous architecture, feature-based methods have been widely validated for their effectiveness. However, in scenarios where the teacher and student models are of heterogeneous architectures, the inherent differences in feature representation significantly degrade the performance of these methods. Recent studies have highlighted that low-frequency components constitute the majority of image features. Motivated by this, we propose a Low-Frequency Components-based Contrastive Knowledge Distillation (LFCC) framework that significantly enhances the performance of feature-based distillation between heterogeneous architectures. Specifically, we designe a set of multi-scale low-pass filters to extract the low-frequency components of intermediate features from both the teacher and student models, aligning them in a compact space to overcome architectural disparities. Moreover, leveraging the intrinsic pairing characteristic of the teacher-student framework, we design an innovative sample-level contrastive learning framework that adeptly restructures the constraints of within-sample feature similarity and between-sample feature divergence into a contrastive learning task. This strategy enables the student model to capitalize on intra-sample feature congruence while simultaneously enhancing the discrimination of features among disparate samples. Consequently, our LFCC framework accurately captures the commonalities in feature representation across heterogeneous architectures. Extensive evaluations and empirical analyses across three architectures (CNNs, Transformers, and MLPs) demonstrate that LFCC achieves superior performance on the challenging benchmarks of ImageNet-1K and CIFAR-100. All codes will be publicly available.

cs.CV↗
arXiv2024

SAMDA: Leveraging SAM on Few-Shot Domain Adaptation for Electronic Microscopy Segmentation

It has been shown that traditional deep learning methods for electronic microscopy segmentation usually suffer from low transferability when samples and annotations are limited, while large-scale vision foundation models are more robust when transferring between different domains but facing sub-optimal improvement under fine-tuning. In this work, we present a new few-shot domain adaptation framework SAMDA, which combines the Segment Anything Model(SAM) with nnUNet in the embedding space to achieve high transferability and accuracy. Specifically, we choose the Unet-based network as the "expert" component to learn segmentation features efficiently and design a SAM-based adaptation module as the "generic" component for domain transfer. By amalgamating the "generic" and "expert" components, we mitigate the modality imbalance in the complex pre-training knowledge inherent to large-scale Vision Foundation models and the challenge of transferability inherent to traditional neural networks. The effectiveness of our model is evaluated on two electron microscopic image datasets with different modalities for mitochondria segmentation, which improves the dice coefficient on the target domain by 6.7%. Also, the SAM-based adaptor performs significantly better with only a single annotated image than the 10-shot domain adaptation on nnUNet. We further verify our model on four MRI datasets from different sources to prove its generalization ability.

eess.IV↗
arXiv2024

Exploring General Intelligence via Gated Graph Transformer in Functional Connectivity Studies

Functional connectivity (FC) as derived from fMRI has emerged as a pivotal tool in elucidating the intricacies of various psychiatric disorders and delineating the neural pathways that underpin cognitive and behavioral dynamics inherent to the human brain. While Graph Neural Networks (GNNs) offer a structured approach to represent neuroimaging data, they are limited by their need for a predefined graph structure to depict associations between brain regions, a detail not solely provided by FCs. To bridge this gap, we introduce the Gated Graph Transformer (GGT) framework, designed to predict cognitive metrics based on FCs. Empirical validation on the Philadelphia Neurodevelopmental Cohort (PNC) underscores the superior predictive prowess of our model, further accentuating its potential in identifying pivotal neural connectivities that correlate with human cognitive processes.

q-bio.NC↗