arXiv subjects

Lin Li

Publications and source records attributed to Lin Li.

At least 163 records · Page 9Linked to original sources

arXiv2025

Pangu Ultra MoE: How to Train Your Big MoE on Ascend NPUs

Sparse large language models (LLMs) with Mixture of Experts (MoE) and close to a trillion parameters are dominating the realm of most capable language models. However, the massive model scale poses significant challenges for the underlying software and hardware systems. In this paper, we aim to uncover a recipe to harness such scale on Ascend NPUs. The key goals are better usage of the computing resources under the dynamic sparse model structures and materializing the expected performance gain on the actual hardware. To select model configurations suitable for Ascend NPUs without repeatedly running the expensive experiments, we leverage simulation to compare the trade-off of various model hyperparameters. This study led to Pangu Ultra MoE, a sparse LLM with 718 billion parameters, and we conducted experiments on the model to verify the simulation results. On the system side, we dig into Expert Parallelism to optimize the communication between NPU devices to reduce the synchronization overhead. We also optimize the memory efficiency within the devices to further reduce the parameter and activation management overhead. In the end, we achieve an MFU of 30.0% when training Pangu Ultra MoE, with performance comparable to that of DeepSeek R1, on 6K Ascend NPUs, and demonstrate that the Ascend system is capable of harnessing all the training stages of the state-of-the-art language models. Extensive experiments indicate that our recipe can lead to efficient training of large-scale sparse language models with MoE. We also study the behaviors of such models for future reference.

cs.CL↗

arXiv2025

Breast density in MRI: an AI-based quantification and relationship to assessment in mammography

Mammographic breast density is a well-established risk factor for breast cancer. Recently there has been interest in breast MRI as an adjunct to mammography, as this modality provides an orthogonal and highly quantitative assessment of breast tissue. However, its 3D nature poses analytic challenges related to delineating and aggregating complex structures across slices. Here, we applied an in-house machine-learning algorithm to assess breast density on normal breasts in three MRI datasets. Breast density was consistent across different datasets (0.104 - 0.114). Analysis across different age groups also demonstrated strong consistency across datasets and confirmed a trend of decreasing density with age as reported in previous studies. MR breast density was correlated with mammographic breast density, although some notable differences suggest that certain breast density components are captured only on MRI. Future work will determine how to integrate MR breast density with current tools to improve future breast cancer risk prediction.

cs.CV↗

arXiv2025

Knowledge Acquisition on Mass-shooting Events via LLMs for AI-Driven Justice

Mass-shooting events pose a significant challenge to public safety, generating large volumes of unstructured textual data that hinder effective investigations and the formulation of public policy. Despite the urgency, few prior studies have effectively automated the extraction of key information from these events to support legal and investigative efforts. This paper presented the first dataset designed for knowledge acquisition on mass-shooting events through the application of named entity recognition (NER) techniques. It focuses on identifying key entities such as offenders, victims, locations, and criminal instruments, that are vital for legal and investigative purposes. The NER process is powered by Large Language Models (LLMs) using few-shot prompting, facilitating the efficient extraction and organization of critical information from diverse sources, including news articles, police reports, and social media. Experimental results on real-world mass-shooting corpora demonstrate that GPT-4o is the most effective model for mass-shooting NER, achieving the highest Micro Precision, Micro Recall, and Micro F1-scores. Meanwhile, o1-mini delivers competitive performance, making it a resource-efficient alternative for less complex NER tasks. It is also observed that increasing the shot count enhances the performance of all models, but the gains are more substantial for GPT-4o and o1-mini, highlighting their superior adaptability to few-shot learning scenarios.

cs.CY↗

arXiv2025

A spectral Levenberg-Marquardt-Deflation method for multiple solutions of semilinear elliptic systems

Many nonlinear differential equations arising from practical problems may permit nontrivial multiple solutions relevant to applications, and these multiple solutions are helpful to deeply understand these practical problems and to improve some applications. Developing an efficient numerical method for finding multiple solutions is very necessary due to the nonlinearity and multiple solutions of these equations. Moreover, providing an efficient iteration plays an important role in successfully obtaining multiple solutions with fast and stable convergence. In the current paper, an efficient algorithm for finding multiple solutions of semilinear elliptic systems is proposed, where the trust region Levenberg-Marquardt method is firstly used to iterate the resulted nonlinear algebraic system. When the nonlinear term in these equations has only the first derivative, our algorithm can efficiently find multiple solutions as well. Several numerical experiments are tested to show the efficiency of our algorithm, and some solutions which have not been shown in the literature are also found and shown.

math.OC↗

arXiv2025

An Improved Adaptive Orthogonal Basis Deflation Method for Multiple Solutions with Applications to Nonlinear Elliptic Equations in Varying Domains

Multiple solutions are common in various non-convex problems arising from industrial and scientific computing. Nonetheless, understanding the nontrivial solutions' qualitative properties seems limited, partially due to the lack of efficient and reliable numerical methods. In this paper, we design a dedicated numerical method to explore these nontrivial solutions further. We first design an improved adaptive orthogonal basis deflation method by combining the adaptive orthogonal basis method with a bisection-deflation algorithm. We then apply the proposed new method to study the impact of domain changes on multiple solutions of certain nonlinear elliptic equations. When the domain varies from a circular disk to an elliptical disk, the corresponding functional value changes dramatically for some particular solutions, which indicates that these nontrivial solutions in the circular domain may become unstable in the elliptical domain. Moreover, several theoretical results on multiple solutions in existing literature are verified. For the nonlinear Sine-Gordon equation with parameter $λ$, nontrivial solutions are found for $λ> λ_2$, here $λ_2$ is the second eigenvalue of the corresponding linear eigenvalue problem. For the singularly perturbed Ginzburg-Landau equation, highly concentrated solutions are numerically found which suggests that their convergent limit is a delta function when the perturbation parameter goes to zero

math.NA↗

arXiv2025

A novel numerical method tailored for unconstrained optimization problems

Unconstrained optimization problems become more common in scientific computing and engineering applications with the rapid development of artificial intelligence, and numerical methods for solving them more quickly and efficiently have been getting more attention and research. Moreover, an efficient method to minimize all kinds of objective functions is urgently needed, especially the nonsmooth objective function. Therefore, in the current paper, we focus on proposing a novel numerical method tailored for unconstrained optimization problems whether the objective function is smooth or not. To be specific, based on the variational procedure to refine the gradient and Hessian matrix approximations, an efficient quadratic model with $2n$ constrained conditions is established. Moreover, to improve the computational efficiency, a simplified model with 2 constrained conditions is also proposed, where the gradient and Hessian matrix can be explicitly updated, and the corresponding boundedness of the remaining $2n-2$ constrained conditions is derived. On the other hand, the novel numerical method is summarized, and approximation results on derivative information are also analyzed and shown. Numerical experiments involving smooth, derivative blasting, and non-smooth problems are tested, demonstrating its feasibility and efficiency. Compared with existing methods, our proposed method can efficiently solve smooth and non-smooth unconstrained optimization problems for the first time, and it is very easy to program the code, indicating that our proposed method not also has great application prospects, but is also very meaningful to explore practical complex engineering and scientific problems.

math.OC↗

arXiv2025

Predicting ulcer in H&E images of inflammatory bowel disease using domain-knowledge-driven graph neural network

Inflammatory bowel disease (IBD) involves chronic inflammation of the digestive tract, with treatment options often burdened by adverse effects. Identifying biomarkers for personalized treatment is crucial. While immune cells play a key role in IBD, accurately identifying ulcer regions in whole slide images (WSIs) is essential for characterizing these cells and exploring potential therapeutics. Multiple instance learning (MIL) approaches have advanced WSI analysis but they lack spatial context awareness. In this work, we propose a weakly-supervised model called DomainGCN that employs a graph convolution neural network (GCN) and incorporates domain-specific knowledge of ulcer features, specifically, the presence of epithelium, lymphocytes, and debris for WSI-level ulcer prediction in IBD. We demonstrate that DomainGCN outperforms various state-of-the-art (SOTA) MIL methods and show the added value of domain knowledge.

eess.IV↗

arXiv2025

Pangu Ultra: Pushing the Limits of Dense Large Language Models on Ascend NPUs

We present Pangu Ultra, a Large Language Model (LLM) with 135 billion parameters and dense Transformer modules trained on Ascend Neural Processing Units (NPUs). Although the field of LLM has been witnessing unprecedented advances in pushing the scale and capability of LLM in recent years, training such a large-scale model still involves significant optimization and system challenges. To stabilize the training process, we propose depth-scaled sandwich normalization, which effectively eliminates loss spikes during the training process of deep models. We pre-train our model on 13.2 trillion diverse and high-quality tokens and further enhance its reasoning capabilities during post-training. To perform such large-scale training efficiently, we utilize 8,192 Ascend NPUs with a series of system optimizations. Evaluations on multiple diverse benchmarks indicate that Pangu Ultra significantly advances the state-of-the-art capabilities of dense LLMs such as Llama 405B and Mistral Large 2, and even achieves competitive results with DeepSeek-R1, whose sparse model structure contains much more parameters. Our exploration demonstrates that Ascend NPUs are capable of efficiently and effectively training dense models with more than 100 billion parameters. Our model and system will be available for our commercial customers.

cs.CL↗

arXiv2025

CoMM: A Coherent Interleaved Image-Text Dataset for Multimodal Understanding and Generation

Interleaved image-text generation has emerged as a crucial multimodal task, aiming at creating sequences of interleaved visual and textual content given a query. Despite notable advancements in recent multimodal large language models (MLLMs), generating integrated image-text sequences that exhibit narrative coherence and entity and style consistency remains challenging due to poor training data quality. To address this gap, we introduce CoMM, a high-quality Coherent interleaved image-text MultiModal dataset designed to enhance the coherence, consistency, and alignment of generated multimodal content. Initially, CoMM harnesses raw data from diverse sources, focusing on instructional content and visual storytelling, establishing a foundation for coherent and consistent content. To further refine the data quality, we devise a multi-perspective filter strategy that leverages advanced pre-trained models to ensure the development of sentences, consistency of inserted images, and semantic alignment between them. Various quality evaluation metrics are designed to prove the high quality of the filtered dataset. Meanwhile, extensive few-shot experiments on various downstream tasks demonstrate CoMM's effectiveness in significantly enhancing the in-context learning capabilities of MLLMs. Moreover, we propose four new tasks to evaluate MLLMs' interleaved generation abilities, supported by a comprehensive evaluation framework. We believe CoMM opens a new avenue for advanced MLLMs with superior multimodal in-context learning and understanding ability.

cs.CV↗

arXiv2025

Emerging Cyber Attack Risks of Medical AI Agents

Large language models (LLMs)-powered AI agents exhibit a high level of autonomy in addressing medical and healthcare challenges. With the ability to access various tools, they can operate within an open-ended action space. However, with the increase in autonomy and ability, unforeseen risks also arise. In this work, we investigated one particular risk, i.e., cyber attack vulnerability of medical AI agents, as agents have access to the Internet through web browsing tools. We revealed that through adversarial prompts embedded on webpages, cyberattackers can: i) inject false information into the agent's response; ii) they can force the agent to manipulate recommendation (e.g., healthcare products and services); iii) the attacker can also steal historical conversations between the user and agent, resulting in the leak of sensitive/private medical information; iv) furthermore, the targeted agent can also cause a computer system hijack by returning a malicious URL in its response. Different backbone LLMs were examined, and we found such cyber attacks can succeed in agents powered by most mainstream LLMs, with the reasoning models such as DeepSeek-R1 being the most vulnerable.

cs.CR↗

arXiv2025

Accelerating and enhancing thermodynamic simulations of electrochemical interfaces

Electrochemical interfaces are crucial in catalysis, energy storage, and corrosion, where their stability and reactivity depend on complex interactions between the electrode, adsorbates, and electrolyte. Predicting stable surface structures remains challenging, as traditional surface Pourbaix diagrams tend to either rely on expert knowledge or costly $\textit{ab initio}$ sampling, and neglect thermodynamic equilibration with the environment. Machine learning (ML) potentials can accelerate static modeling but often overlook dynamic surface transformations. Here, we extend the Virtual Surface Site Relaxation-Monte Carlo (VSSR-MC) method to autonomously sample surface reconstructions modeled under aqueous electrochemical conditions. Through fine-tuning foundational ML force fields, we accurately and efficiently predict surface energetics, recovering known Pt(111) phases and revealing new LaMnO$_\mathrm{3}$(001) surface reconstructions. By explicitly accounting for bulk-electrolyte equilibria, our framework enhances electrochemical stability predictions, offering a scalable approach to understanding and designing materials for electrochemical applications.

cond-mat.mtrl-sci↗

arXiv2025

Personalize Anything for Free with Diffusion Transformer

Personalized image generation aims to produce images of user-specified concepts while enabling flexible editing. Recent training-free approaches, while exhibit higher computational efficiency than training-based methods, struggle with identity preservation, applicability, and compatibility with diffusion transformers (DiTs). In this paper, we uncover the untapped potential of DiT, where simply replacing denoising tokens with those of a reference subject achieves zero-shot subject reconstruction. This simple yet effective feature injection technique unlocks diverse scenarios, from personalization to image editing. Building upon this observation, we propose \textbf{Personalize Anything}, a training-free framework that achieves personalized image generation in DiT through: 1) timestep-adaptive token replacement that enforces subject consistency via early-stage injection and enhances flexibility through late-stage regularization, and 2) patch perturbation strategies to boost structural diversity. Our method seamlessly supports layout-guided generation, multi-subject personalization, and mask-controlled editing. Evaluations demonstrate state-of-the-art performance in identity preservation and versatility. Our work establishes new insights into DiTs while delivering a practical paradigm for efficient personalization.

cs.CV↗

arXiv2025

A Novel Ophthalmic Benchmark for Evaluating Multimodal Large Language Models with Fundus Photographs and OCT Images

In recent years, large language models (LLMs) have demonstrated remarkable potential across various medical applications. Building on this foundation, multimodal large language models (MLLMs) integrate LLMs with visual models to process diverse inputs, including clinical data and medical images. In ophthalmology, LLMs have been explored for analyzing optical coherence tomography (OCT) reports, assisting in disease classification, and even predicting treatment outcomes. However, existing MLLM benchmarks often fail to capture the complexities of real-world clinical practice, particularly in the analysis of OCT images. Many suffer from limitations such as small sample sizes, a lack of diverse OCT datasets, and insufficient expert validation. These shortcomings hinder the accurate assessment of MLLMs' ability to interpret OCT scans and their broader applicability in ophthalmology. Our dataset, curated through rigorous quality control and expert annotation, consists of 439 fundus images and 75 OCT images. Using a standardized API-based framework, we assessed seven mainstream MLLMs and observed significant variability in diagnostic accuracy across different diseases. While some models performed well in diagnosing conditions such as diabetic retinopathy and age-related macular degeneration, they struggled with others, including choroidal neovascularization and myopia, highlighting inconsistencies in performance and the need for further refinement. Our findings emphasize the importance of developing clinically relevant benchmarks to provide a more accurate assessment of MLLMs' capabilities. By refining these models and expanding their scope, we can enhance their potential to transform ophthalmic diagnosis and treatment.

cs.CL↗

arXiv2025

Reducing Large Language Model Safety Risks in Women's Health using Semantic Entropy

Large language models (LLMs) hold substantial promise for clinical decision support. However, their widespread adoption in medicine, particularly in healthcare, is hindered by their propensity to generate false or misleading outputs, known as hallucinations. In high-stakes domains such as women's health (obstetrics & gynaecology), where errors in clinical reasoning can have profound consequences for maternal and neonatal outcomes, ensuring the reliability of AI-generated responses is critical. Traditional methods for quantifying uncertainty, such as perplexity, fail to capture meaning-level inconsistencies that lead to misinformation. Here, we evaluate semantic entropy (SE), a novel uncertainty metric that assesses meaning-level variation, to detect hallucinations in AI-generated medical content. Using a clinically validated dataset derived from UK RCOG MRCOG examinations, we compared SE with perplexity in identifying uncertain responses. SE demonstrated superior performance, achieving an AUROC of 0.76 (95% CI: 0.75-0.78), compared to 0.62 (0.60-0.65) for perplexity. Clinical expert validation further confirmed its effectiveness, with SE achieving near-perfect uncertainty discrimination (AUROC: 0.97). While semantic clustering was successful in only 30% of cases, SE remains a valuable tool for improving AI safety in women's health. These findings suggest that SE could enable more reliable AI integration into clinical practice, particularly in resource-limited settings where LLMs could augment care. This study highlights the potential of SE as a key safeguard in the responsible deployment of AI-driven tools in women's health, leading to safer and more effective digital health interventions.

cs.LG↗

arXiv2025

Random Forest-of-Thoughts: Uncertainty-aware Reasoning for Computational Social Science

Social surveys in computational social science are well-designed by elaborate domain theories that can effectively reflect the interviewee's deep thoughts without concealing their true feelings. The candidate questionnaire options highly depend on the interviewee's previous answer, which results in the complexity of social survey analysis, the time, and the expertise required. The ability of large language models (LLMs) to perform complex reasoning is well-enhanced by prompting learning such as Chain-of-thought (CoT) but still confined to left-to-right decision-making processes or limited paths during inference. This means they can fall short in problems that require exploration and uncertainty searching. In response, a novel large language model prompting method, called Random Forest of Thoughts (RFoT), is proposed for generating uncertainty reasoning to fit the area of computational social science. The RFoT allows LLMs to perform deliberate decision-making by generating diverse thought space and randomly selecting the sub-thoughts to build the forest of thoughts. It can extend the exploration and prediction of overall performance, benefiting from the extensive research space of response. The method is applied to optimize computational social science analysis on two datasets covering a spectrum of social survey analysis problems. Our experiments show that RFoT significantly enhances language models' abilities on two novel social survey analysis problems requiring non-trivial reasoning.

cs.CL↗

arXiv2025

Exploring Incremental Unlearning: Techniques, Challenges, and Future Directions

The growing demand for data privacy in Machine Learning (ML) applications has seen Machine Unlearning (MU) emerge as a critical area of research. As the `right to be forgotten' becomes regulated globally, it is increasingly important to develop mechanisms that delete user data from AI systems while maintaining performance and scalability of these systems. Incremental Unlearning (IU) is a promising MU solution to address the challenges of efficiently removing specific data from ML models without the need for expensive and time-consuming full retraining. This paper presents the various techniques and approaches to IU. It explores the challenges faced in designing and implementing IU mechanisms. Datasets and metrics for evaluating the performance of unlearning techniques are discussed as well. Finally, potential solutions to the IU challenges alongside future research directions are offered. This survey provides valuable insights for researchers and practitioners seeking to understand the current landscape of IU and its potential for enhancing privacy-preserving intelligent systems.

cs.LG↗

arXiv2025

The neutrino luminosity and energy spectrum of nova outburst

The nova outburst can produce a large number of neutrinos, whether it is the nuclear reaction process during the explosion or the shock wave acceleration proton process. We study the low-energy nuclear and thermal neutrino luminosity of novae with CO white dwarf (WD) mass ranging from 0.6 to 1.1 $\rm M_{\odot}$ with different accretion rates $\dot{M}$, core temperatures $(T_{\mathrm{C}})$, and mixing degrees. We find that during the accretion phase, low-energy neutrinos are mainly produced by pp chains and plasma decay, and photon luminosity is greater than low-energy nuclear and thermal neutrino luminosity. During the thermonuclear runaway (TNR) phase, low-energy neutrinos are mainly produced by the CNO cycle and photon-neutrino, and the low-energy nuclear and thermal neutrino luminosity far exceeds the photon luminosity. We find that the more massive the WD, the shorter the cycle time and the higher the low-energy nuclear neutrino luminosity. The higher the accretion rate, the lower the low-energy nuclear neutrino luminosity. If the accretion mixing effect is not taken into account, the outburst interval becomes longer, the low-energy nuclear neutrino luminosity will be increased. And for the cooler nova model $(T_{\mathrm{C}}=1\times10^{7}\rm K)$, the low-energy nuclear neutrino luminosity will be lower during the accretion phase and higher at the TNR. We also predict the neutrino luminosity and energy spectrum of the upcoming recurrent nova T Coronae Borealis (T CrB). We estimate that the next T CrB outburst has a low-energy nuclear neutrino peak luminosity of $2.7\times10^{8}\ \rm L_{ν,\odot}$ and a low-energy nuclear neutrino outburst duration of 88 days. In addition, we predict that the high-energy hadronic neutrino flux produced by T CrB nova can not be observed by the current-generation IceCube.

astro-ph.SR↗

arXiv2025

Taking A Closer Look at Interacting Objects: Interaction-Aware Open Vocabulary Scene Graph Generation

Today's open vocabulary scene graph generation (OVSGG) extends traditional SGG by recognizing novel objects and relationships beyond predefined categories, leveraging the knowledge from pre-trained large-scale models. Most existing methods adopt a two-stage pipeline: weakly supervised pre-training with image captions and supervised fine-tuning (SFT) on fully annotated scene graphs. Nonetheless, they omit explicit modeling of interacting objects and treat all objects equally, resulting in mismatched relation pairs. To this end, we propose an interaction-aware OVSGG framework INOVA. During pre-training, INOVA employs an interaction-aware target generation strategy to distinguish interacting objects from non-interacting ones. In SFT, INOVA devises an interaction-guided query selection tactic to prioritize interacting objects during bipartite graph matching. Besides, INOVA is equipped with an interaction-consistent knowledge distillation to enhance the robustness by pushing interacting object pairs away from the background. Extensive experiments on two benchmarks (VG and GQA) show that INOVA achieves state-of-the-art performance, demonstrating the potential of interaction-aware mechanisms for real-world applications.

cs.CV↗