arXiv subjects

Yan Li

Publications and source records attributed to Yan Li.

At least 181 records · Page 10Linked to original sources

arXiv2025

Centrum: Model-based Database Auto-tuning with Minimal Distributional Assumptions

Gaussian-Process-based Bayesian optimization (GP-BO), is a prevailing model-based framework for DBMS auto-tuning. However, recent work shows GP-BO-based DBMS auto-tuners significantly outperformed auto-tuners based on SMAC, which features random forest surrogate models; such results motivate us to rethink and investigate the limitations of GP-BO in auto-tuner design. We find the fundamental assumptions of GP-BO are widely violated when modeling and optimizing DBMS performance, while tree-ensemble-BOs (e.g., SMAC) can avoid the assumption pitfalls and deliver improved tuning efficiency and effectiveness. Moreover, we argue that existing tree-ensemble-BOs restrict further advancement in DBMS auto-tuning. First, existing tree-ensemble-BOs can only achieve distribution-free point estimates, but still impose unrealistic distributional assumptions on uncertainty estimates, compromising surrogate modeling and distort the acquisition function. Second, recent advances in gradient boosting, which can further enhance surrogate modeling against vanilla GP and random forest counterparts, have rarely been applied in optimizing DBMS auto-tuners. To address these issues, we propose a novel model-based DBMS auto-tuner, Centrum. Centrum improves distribution-free point and interval estimation in surrogate modeling with a two-phase learning procedure of stochastic gradient boosting ensembles. Moreover, Centrum adopts a generalized SGBE-estimated locally-adaptive conformal prediction to facilitate a distribution-free uncertainty estimation and acquisition function. To our knowledge, Centrum is the first auto-tuner to realize distribution-freeness, enhancing BO's practicality in DBMS auto-tuning, and the first to seamlessly fuse gradient boosting ensembles and conformal inference in BO. Extensive physical and simulation experiments on two DBMSs and three workloads show Centrum outperforms 21 SOTA methods.

cs.LG↗

arXiv2025

The oscillation properties of the Blue Large Amplitude Pulsators (BLAPs): relative change rate of periods, excitations, and period relations

\textbf{B}lue \textbf{L}arge \textbf{A}mplitude \textbf{P}ulsators (BLAPs) are a type of variable star that has been identified relatively recently. They are characterized by their large amplitude, high gravity, and long periods (2-75 minutes). Some BLAPs exhibit a rich helium abundance on their surfaces, while some of the others show rich hydrogen atmospheres. Up to date, both the mode identification and formation pathways of BLAPs remain subjects of ongoing debate. In this work, we gave an non-adiabatic pulsation analyses for low-order radial and non-radial dipole modes on a theoretical model grid to explore the characterist ics of BLAPs. The effects of element diffusion and radiative levitation are not taken into account our theoretical models. This because that these processes could lead to discrepancies in surface element abundance of the sdB or BLAP stars between theoretical models and observations. Based on the theoretical model grid, our analyses revealed that the radial fundamental modes align well with various observed properties of BLAPs, including the pulsation instability, the position of the excitable models on the HR diagram, surface element compositions, period ranges, and period changes. Our theoretical models also predicted a period-gravity ($\log g$) relation with a narrow distribution, while the observations exhibit a broader range. Furthermore, our findings indicate that the excitation of pulsation in BLAPs is primarily influenced by metal abundance. The thickness of the outer rich hydrogen envelope and the ratio of helium to hydrogen on the surface almost do not affect the pulsation instability.

astro-ph.SR↗

arXiv2025

Uniform K-stability of $G$-varieties of complexity 1

Let ${\rm k}$ be an algebraically closed field of characteristic 0 and $G$ a connect, reductive group over it. Let $X$ be a projective $G$-variety of complexity 1. We classify $G$-equivariant normal test configurations of $X$ with integral central fibre via the combinatorial data. We also give a formula of anti-canonical divisors on $X$. Based on this formula, when $X$ is $\mathbb Q$-Fano, we give an expression of the Futaki invariant, and derive a criterion of uniform K-stability in terms of the combinatorial data.

math.AG↗

arXiv2025

Unity is Power: Semi-Asynchronous Collaborative Training of Large-Scale Models with Structured Pruning in Resource-Limited Clients

In this work, we study to release the potential of massive heterogeneous weak computing power to collaboratively train large-scale models on dispersed datasets. In order to improve both efficiency and accuracy in resource-adaptive collaborative learning, we take the first step to consider the \textit{unstructured pruning}, \textit{varying submodel architectures}, \textit{knowledge loss}, and \textit{straggler} challenges simultaneously. We propose a novel semi-asynchronous collaborative training framework, namely ${Co\text{-}S}^2{P}$, with data distribution-aware structured pruning and cross-block knowledge transfer mechanism to address the above concerns. Furthermore, we provide theoretical proof that ${Co\text{-}S}^2{P}$ can achieve asymptotic optimal convergence rate of $O(1/\sqrt{N^*EQ})$. Finally, we conduct extensive experiments on two types of tasks with a real-world hardware testbed including diverse IoT devices.The experimental results demonstrate that $Co\text{-}S^2P$ improves accuracy by up to 8.8\% and resource utilization by up to 1.2$\times$ compared to state-of-the-art methods, while reducing memory consumption by approximately 22\% and training time by about 24\% on all resource-limited devices.

cs.DC↗

arXiv2025

CausalVerse: Benchmarking Causal Representation Learning with Configurable High-Fidelity Simulations

Causal Representation Learning (CRL) aims to uncover the data-generating process and identify the underlying causal variables and relations, whose evaluation remains inherently challenging due to the requirement of known ground-truth causal variables and causal structure. Existing evaluations often rely on either simplistic synthetic datasets or downstream performance on real-world tasks, generally suffering a dilemma between realism and evaluative precision. In this paper, we introduce a new benchmark for CRL using high-fidelity simulated visual data that retains both realistic visual complexity and, more importantly, access to ground-truth causal generating processes. The dataset comprises around 200 thousand images and 3 million video frames across 24 sub-scenes in four domains: static image generation, dynamic physical simulations, robotic manipulations, and traffic situation analysis. These scenarios range from static to dynamic settings, simple to complex structures, and single to multi-agent interactions, offering a comprehensive testbed that hopefully bridges the gap between rigorous evaluation and real-world applicability. In addition, we provide flexible access to the underlying causal structures, allowing users to modify or configure them to align with the required assumptions in CRL, such as available domain labels, temporal dependencies, or intervention histories. Leveraging this benchmark, we evaluated representative CRL methods across diverse paradigms and offered empirical insights to assist practitioners and newcomers in choosing or extending appropriate CRL frameworks to properly address specific types of real problems that can benefit from the CRL perspective. Welcome to visit our: Project page:https://causal-verse.github.io/, Dataset:https://huggingface.co/CausalVerse.

cs.LG↗

arXiv2025

Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space

We propose AdapTok, an adaptive temporal causal video tokenizer that can flexibly allocate tokens for different frames based on video content. AdapTok is equipped with a block-wise masking strategy that randomly drops tail tokens of each block during training, and a block causal scorer to predict the reconstruction quality of video frames using different numbers of tokens. During inference, an adaptive token allocation strategy based on integer linear programming is further proposed to adjust token usage given predicted scores. Such design allows for sample-wise, content-aware, and temporally dynamic token allocation under a controllable overall budget. Extensive experiments for video reconstruction and generation on UCF-101 and Kinetics-600 demonstrate the effectiveness of our approach. Without additional image data, AdapTok consistently improves reconstruction quality and generation performance under different token budgets, allowing for more scalable and token-efficient generative video modeling.

cs.CV↗

arXiv2025

L2M-AID: Autonomous Cyber-Physical Defense by Fusing Semantic Reasoning of Large Language Models with Multi-Agent Reinforcement Learning (Preprint)

The increasing integration of Industrial IoT (IIoT) exposes critical cyber-physical systems to sophisticated, multi-stage attacks that elude traditional defenses lacking contextual awareness. This paper introduces L2M-AID, a novel framework for Autonomous Industrial Defense using LLM-empowered, Multi-agent reinforcement learning. L2M-AID orchestrates a team of collaborative agents, each driven by a Large Language Model (LLM), to achieve adaptive and resilient security. The core innovation lies in the deep fusion of two AI paradigms: we leverage an LLM as a semantic bridge to translate vast, unstructured telemetry into a rich, contextual state representation, enabling agents to reason about adversary intent rather than merely matching patterns. This semantically-aware state empowers a Multi-Agent Reinforcement Learning (MARL) algorithm, MAPPO, to learn complex cooperative strategies. The MARL reward function is uniquely engineered to balance security objectives (threat neutralization) with operational imperatives, explicitly penalizing actions that disrupt physical process stability. To validate our approach, we conduct extensive experiments on the benchmark SWaT dataset and a novel synthetic dataset generated based on the MITRE ATT&CK for ICS framework. Results demonstrate that L2M-AID significantly outperforms traditional IDS, deep learning anomaly detectors, and single-agent RL baselines across key metrics, achieving a 97.2% detection rate while reducing false positives by over 80% and improving response times by a factor of four. Crucially, it demonstrates superior performance in maintaining physical process stability, presenting a robust new paradigm for securing critical national infrastructure.

cs.AI↗

arXiv2025

Recent Advances in Microfluidics and Bioelectronics for Three-Dimensional Organoid Interfaces

Organoids offer a promising alternative in biomedical research and clinical medicine, with better feature recapitulation than 2D cultures. They also have more consistent responses with clinical results when compared to animal models. However, major challenges exist in the longevity of culture, the reproducibility of organoid properties, and the development of non-disruptive monitoring methods. Recent advances in materials and microfabrication methods, such as 3D printing and compressive buckling, have enabled three-dimensional (3D) interfaces of microfluidics and bioelectronics to manipulate and monitor these biological models in exciting ways. These advanced systems have great potential for applications in drug delivery, personalized medicine, and disease modelling. We conclude with important future considerations to generate longevity using further technological development in organoid and spheroid models.

q-bio.TO↗

arXiv2025

Towards Self-Refinement of Vision-Language Models with Triangular Consistency

Vision-Language Models (VLMs) integrate visual knowledge with the analytical capabilities of Large Language Models (LLMs) through supervised visual instruction tuning, using image-question-answer triplets. However, the potential of VLMs trained without supervised instruction remains largely unexplored. This study validates that VLMs possess inherent self-refinement capabilities, enabling them to generate high-quality supervised data without external inputs and thereby learn autonomously. Specifically, to stimulate the self-refinement ability of VLMs, we propose a self-refinement framework based on a Triangular Consistency principle: within the image-query-answer triangle, any masked elements should be consistently and accurately reconstructed. The framework involves three steps: (1) We enable the instruction generation ability of VLMs by adding multi-task instruction tuning like image$\rightarrow$question-answer or image-answer$\rightarrow$question. (2) We generate image-query-answer triplets from unlabeled images and use the Triangular Consistency principle for filtering. (3) The model is further updated using the filtered synthetic data. To investigate the underlying mechanisms behind this self-refinement capability, we conduct a theoretical analysis from a causal perspective. Using the widely recognized LLaVA-1.5 as our baseline, our experiments reveal that the model can autonomously achieve consistent, though deliberately modest, improvements across multiple benchmarks without any external supervision, such as human annotations or environmental feedback. We expect that the insights of this study on the self-refinement ability of VLMs can inspire future research on the learning mechanism of VLMs. Code is available at https://github.com/dengyl20/SRF-LLaVA-1.5.

cs.CV↗

arXiv2025

Cluster-Aware Prompt Ensemble Learning for Few-Shot Vision-Language Model Adaptation

Vision-language models (VLMs) such as CLIP achieve zero-shot transfer across various tasks by pre-training on numerous image-text pairs. These models often benefit from using an ensemble of context prompts to represent a class. Despite being effective, conventional prompt ensembling that averages textual features of context prompts often yields suboptimal results. This is because feature averaging shifts the class centroids away from the true class distribution. To address this issue, we propose the Cluster-Aware Prompt Ensemble Learning (CAPEL) framework, which preserves the cluster nature of context prompts. CAPEL classifies images into one of several class clusters, each represented by a distinct prompt. Instead of ensembling prompts in the feature space, we perform ensembling in the classification logits space, aligning better with the visual feature distribution. To further optimize prompt fine-tuning while maintaining cluster-specific discriminative power, we introduce a cluster-preserving regularization term. This ensures that prompts remain distinct and specialized for different clusters, preventing collapse into a uniform direction. Additionally, we integrate an adaptive prompt weighting technique to dynamically adjust the attention weights for flawed or ambiguous prompts, ensuring robust performance across diverse datasets and tasks.

cs.CV↗

arXiv2025

When Semantics Mislead Vision: Mitigating Large Multimodal Models Hallucinations in Scene Text Spotting and Understanding

Large Multimodal Models (LMMs) have achieved impressive progress in visual perception and reasoning. However, when confronted with visually ambiguous or non-semantic scene text, they often struggle to accurately spot and understand the content, frequently generating semantically plausible yet visually incorrect answers, which we refer to as semantic hallucination. In this work, we investigate the underlying causes of semantic hallucination and identify a key finding: Transformer layers in LLM with stronger attention focus on scene text regions are less prone to producing semantic hallucinations. Thus, we propose a training-free semantic hallucination mitigation framework comprising two key components: (1) ZoomText, a coarse-to-fine strategy that identifies potential text regions without external detectors; and (2) Grounded Layer Correction, which adaptively leverages the internal representations from layers less prone to hallucination to guide decoding, correcting hallucinated outputs for non-semantic samples while preserving the semantics of meaningful ones. To enable rigorous evaluation, we introduce TextHalu-Bench, a benchmark of 1,740 samples spanning both semantic and non-semantic cases, with manually curated question answer pairs designed to probe model hallucinations. Extensive experiments demonstrate that our method not only effectively mitigates semantic hallucination but also achieves strong performance on public benchmarks for scene text spotting and understanding.

cs.CV↗

arXiv2025

Relief of EGFR/FOS-downregulated miR-103a by loganin alleviates NF-kappaB-triggered inflammation and gut barrier disruption in colitis

Due to the ever-rising global incidence rate of inflammatory bowel disease (IBD) and the lack of effective clinical treatment drugs, elucidating the detailed pathogenesis, seeking novel targets, and developing promising drugs are the top priority for IBD treatment. Here, we demonstrate that the levels of microRNA (miR)-103a were significantly downregulated in the inflamed mucosa of ulcerative colitis (UC) patients, along with elevated inflammatory cytokines (IL-1beta/TNF-alpha) and reduced tight junction protein (Occludin/ZO-1) levels, as compared with healthy control objects. Consistently, miR-103a deficient intestinal epithelial cells Caco-2 showed serious inflammatory responses and increased permeability, and DSS induced more severe colitis in miR-103a-/- mice than wild-type ones. Mechanistic studies unraveled that c-FOS suppressed miR-103a transcription via binding to its promoter, then miR-103a-targeted NF-kappaB activation contributes to inflammatory responses and barrier disruption by targeting TAB2 and TAK1. Notably, the traditional Chinese medicine Cornus officinalis (CO) and its core active ingredient loganin potently mitigated inflammation and barrier disruption in UC by specifically blocking the EGFR/RAS/ERK/c-FOS signaling axis, these effects mainly attributed to modulated miR-103a levels as the therapeutic activities of them were almost completely shielded in miR-103a KO mice. Taken together, this work reveals that loganin relieves EGFR/c-FOS axis-suppressed epithelial miR-103a expression, thereby inhibiting NF-kappaB pathway activation, suppressing inflammatory responses, and preserving tight junction integrity in UC. Thus, our data enrich mechanistic insights and promising targets for UC treatment.

q-bio.BM↗

arXiv2025

Finding the temperature window for atomic layer deposition of ruthenium metal via efficient phonon calculations

We investigate the use of first principles thermodynamics based on periodic density functional theory (DFT) to examine the gas-surface chemistry of an oxidized ruthenium surface reacting with hydrogen gas. This reaction system features in the growth of ultrathin Ru films by atomic layer deposition (ALD). We reproduce and rationalize the experimental observation that ALD of the metal from RuO4 and H2 occurs only in a narrow temperature window above 100°C, and this validates the approach. Specifically, the temperature-dependent reaction free energies are computed for the competing potential reactions of the H2 reagent, and show that surface oxide is reduced to water, which is predicted to desorb thermally above 113°C, exposing bare Ru that can further react to surface hydride, and hence deposit Ru metal. The saturating coverages give a predicted growth rate of 0.7 Å/cycle of Ru. At lower temperatures, free energies indicate that water is retained at the surface and reacts with the RuO4 precursor to form an oxide film, also in agreement with experiment. The temperature dependence is obtained with the required accuracy by computing Gibbs free energy corrections from phonon calculations within the harmonic approximation. Surface phonons are computed rapidly and efficiently by parallelization on a cloud architecture within the Schrödinger Materials Science Suite. We also show that rotational and translational entropy of gases dominate the free energies, permitting an alternative approach without phonon calculations, which would be suitable for rapid pre-screening of gas-surface chemistries.

cond-mat.mtrl-sci↗

arXiv2025

Diffusion Model as a Noise-Aware Latent Reward Model for Step-Level Preference Optimization

Preference optimization for diffusion models aims to align them with human preferences for images. Previous methods typically use Vision-Language Models (VLMs) as pixel-level reward models to approximate human preferences. However, when used for step-level preference optimization, these models face challenges in handling noisy images of different timesteps and require complex transformations into pixel space. In this work, we show that pre-trained diffusion models are naturally suited for step-level reward modeling in the noisy latent space, as they are explicitly designed to process latent images at various noise levels. Accordingly, we propose the Latent Reward Model (LRM), which repurposes components of the diffusion model to predict preferences of latent images at arbitrary timesteps. Building on LRM, we introduce Latent Preference Optimization (LPO), a step-level preference optimization method conducted directly in the noisy latent space. Experimental results indicate that LPO significantly improves the model's alignment with general, aesthetic, and text-image alignment preferences, while achieving a 2.5-28x training speedup over existing preference optimization methods. Our code and models are available at https://github.com/Kwai-Kolors/LPO.

cs.CV↗

arXiv2025

Pilot selection in the era of Virtual reality: algorithms for accurate and interpretable machine learning models

With the rapid growth of the aviation industry, there is a need for a large number of flight crew. How to select the right pilots in a cost-efficient manner has become an important research question. In the current study, twenty-three pilots were recruited from China Eastern Airlines, and 23 novices were from the community of Tsinghua University. A novel approach incorporating machine learning and virtual reality technology was applied to distinguish features between these participants with different flight skills. Results indicate that SVM with the MIC feature selection method consistently achieved the highest prediction performance on all metrics with an Accuracy of 0.93, an AUC of 0.96, and an F1 of 0.93, which outperforms four other classifier algorithms and two other feature selection methods. From the perspective of feature selection methods, the MIC method can select features with a nonlinear relationship to sampling labels, instead of a simple filter-out. Our new implementation of the SVM + MIC algorithm outperforms all existing pilot selection algorithms and perhaps provides the first implementation based on eye tracking and flight dynamics data. This study's VR simulation platforms and algorithms can be used for pilot selection and training.

cs.LG↗

arXiv2025

Half-order Fine-Tuning for Diffusion Model: A Recursive Likelihood Ratio Optimizer

The probabilistic diffusion model (DM), generating content by inferencing through a recursive chain structure, has emerged as a powerful framework for visual generation. After pre-training on enormous data, the model needs to be properly aligned to meet requirements for downstream applications. How to efficiently align the foundation DM is a crucial task. Contemporary methods are either based on Reinforcement Learning (RL) or truncated Backpropagation (BP). However, RL and truncated BP suffer from low sample efficiency and biased gradient estimation, respectively, resulting in limited improvement or, even worse, complete training failure. To overcome the challenges, we propose the Recursive Likelihood Ratio (RLR) optimizer, a Half-Order (HO) fine-tuning paradigm for DM. The HO gradient estimator enables the computation graph rearrangement within the recursive diffusive chain, making the RLR's gradient estimator an unbiased one with lower variance than other methods. We theoretically investigate the bias, variance, and convergence of our method. Extensive experiments are conducted on image and video generation to validate the superiority of the RLR. Furthermore, we propose a novel prompt technique that is natural for the RLR to achieve a synergistic effect.

cs.CV↗

arXiv2025

SynDoc: A Hybrid Discriminative-Generative Framework for Enhancing Synthetic Domain-Adaptive Document Key Information Extraction

Domain-specific Visually Rich Document Understanding (VRDU) presents significant challenges due to the complexity and sensitivity of documents in fields such as medicine, finance, and material science. Existing Large (Multimodal) Language Models (LLMs/MLLMs) achieve promising results but face limitations such as hallucinations, inadequate domain adaptation, and reliance on extensive fine-tuning datasets. This paper introduces SynDoc, a novel framework that combines discriminative and generative models to address these challenges. SynDoc employs a robust synthetic data generation workflow, using structural information extraction and domain-specific query generation to produce high-quality annotations. Through adaptive instruction tuning, SynDoc improves the discriminative model's ability to extract domain-specific knowledge. At the same time, a recursive inferencing mechanism iteratively refines the output of both models for stable and accurate predictions. This framework demonstrates scalable, efficient, and precise document understanding and bridges the gap between domain-specific adaptation and general world knowledge for document key information extraction tasks.

cs.CV↗

arXiv2025

Exploiting Block Coordinate Descent for Cost-Effective LLM Model Training

Training large language models typically demands extensive GPU memory and substantial financial investment, which poses a barrier for many small- to medium-sized teams. In this paper, we propose a full-parameter pre-training and fine-tuning framework based on block coordinate descent (BCD), enhanced with engineering optimizations, to enable efficient training of large-scale models on cost-effective RTX 4090, A100 and A800 GPU clusters. Under identical hardware configurations, we reduce the training cost of a 7B model to 33% on A100/A800 and only 2.6% on RTX 4090, compared to standard full-parameter training. It also enables large models previously restricted to A100 clusters to be trained on RTX 4090 without degrading performance. BCD achieves comparable or better accuracy than full-parameter and fine-tuning methods at most cases, with lower GPU consumption and improved hardware utilization.

cs.LG↗