arXiv subjects

Rong Zhou

Publications and source records attributed to Rong Zhou.

At least 37 records · Page 2Linked to original sources

arXiv2025

SAMed-2: Selective Memory Enhanced Medical Segment Anything Model

Recent "segment anything" efforts show promise by learning from large-scale data, but adapting such models directly to medical images remains challenging due to the complexity of medical data, noisy annotations, and continual learning requirements across diverse modalities and anatomical structures. In this work, we propose SAMed-2, a new foundation model for medical image segmentation built upon the SAM-2 architecture. Specifically, we introduce a temporal adapter into the image encoder to capture image correlations and a confidence-driven memory mechanism to store high-certainty features for later retrieval. This memory-based strategy counters the pervasive noise in large-scale medical datasets and mitigates catastrophic forgetting when encountering new tasks or modalities. To train and evaluate SAMed-2, we curate MedBank-100k, a comprehensive dataset spanning seven imaging modalities and 21 medical segmentation tasks. Our experiments on both internal benchmarks and 10 external datasets demonstrate superior performance over state-of-the-art baselines in multi-task scenarios. The code is available at: https://github.com/ZhilingYan/Medical-SAM-Bench.

cs.CV↗

arXiv2025

EfficientLLM: Efficiency in Large Language Models

Large Language Models (LLMs) have driven significant progress, yet their growing parameter counts and context windows incur prohibitive compute, energy, and monetary costs. We introduce EfficientLLM, a novel benchmark and the first comprehensive empirical study evaluating efficiency techniques for LLMs at scale. Conducted on a production-class cluster (48xGH200, 8xH200 GPUs), our study systematically explores three key axes: (1) architecture pretraining (efficient attention variants: MQA, GQA, MLA, NSA; sparse Mixture-of-Experts (MoE)), (2) fine-tuning (parameter-efficient methods: LoRA, RSLoRA, DoRA), and (3) inference (quantization methods: int4, float16). We define six fine-grained metrics (Memory Utilization, Compute Utilization, Latency, Throughput, Energy Consumption, Compression Rate) to capture hardware saturation, latency-throughput balance, and carbon cost. Evaluating over 100 model-technique pairs (0.5B-72B parameters), we derive three core insights: (i) Efficiency involves quantifiable trade-offs: no single method is universally optimal; e.g., MoE reduces FLOPs and improves accuracy but increases VRAM by 40%, while int4 quantization cuts memory/energy by up to 3.9x at a 3-5% accuracy drop. (ii) Optima are task- and scale-dependent: MQA offers optimal memory-latency trade-offs for constrained devices, MLA achieves lowest perplexity for quality-critical tasks, and RSLoRA surpasses LoRA efficiency only beyond 14B parameters. (iii) Techniques generalize across modalities: we extend evaluations to Large Vision Models (Stable Diffusion 3.5, Wan 2.1) and Vision-Language Models (Qwen2.5-VL), confirming effective transferability. By open-sourcing datasets, evaluation pipelines, and leaderboards, EfficientLLM provides essential guidance for researchers and engineers navigating the efficiency-performance landscape of next-generation foundation models.

cs.CL↗

arXiv2025

ELITE: Embedding-Less retrieval with Iterative Text Exploration

Large Language Models (LLMs) have achieved impressive progress in natural language processing, but their limited ability to retain long-term context constrains performance on document-level or multi-turn tasks. Retrieval-Augmented Generation (RAG) mitigates this by retrieving relevant information from an external corpus. However, existing RAG systems often rely on embedding-based retrieval trained on corpus-level semantic similarity, which can lead to retrieving content that is semantically similar in form but misaligned with the question's true intent. Furthermore, recent RAG variants construct graph- or hierarchy-based structures to improve retrieval accuracy, resulting in significant computation and storage overhead. In this paper, we propose an embedding-free retrieval framework. Our method leverages the logical inferencing ability of LLMs in retrieval using iterative search space refinement guided by our novel importance measure and extend our retrieval results with logically related information without explicit graph construction. Experiments on long-context QA benchmarks, including NovelQA and Marathon, show that our approach outperforms strong baselines while reducing storage and runtime by over an order of magnitude.

cs.CL↗

arXiv2025

Strongly compatible systems associated to semistable abelian varieties

We prove a motivic refinement of a result of Weil, Deligne and Raynaud on the existence of strongly compatible systems associated to abelian varieties. More precisely, given an abelian variety $A$ over a number field $\mathrm{E}\subset \mathbb C$, we prove that after replacing $\mathbb E$ by a finite extension, the action of $\mathrm{Gal}(\overline{\mathrm E}/\mathrm E)$ on the $\ell$-adic cohomology $\mathrm H^1_{\mathrm{\acute{e}t}}(A_{\overline{\mathrm E}},\mathbb Q_\ell)$ gives rise to a strongly compatible system of $\ell$-adic representations valued in the Mumford--Tate group $\mathbf G$ of $A$. This involves an independence of $\ell$-statement for the Weil--Deligne representation associated to $A$ at places of semistable reduction, extending previous work of ours at places of good reduction.

math.NT↗

arXiv2025

ClusMFL: A Cluster-Enhanced Framework for Modality-Incomplete Multimodal Federated Learning in Brain Imaging Analysis

Multimodal Federated Learning (MFL) has emerged as a promising approach for collaboratively training multimodal models across distributed clients, particularly in healthcare domains. In the context of brain imaging analysis, modality incompleteness presents a significant challenge, where some institutions may lack specific imaging modalities (e.g., PET, MRI, or CT) due to privacy concerns, device limitations, or data availability issues. While existing work typically assumes modality completeness or oversimplifies missing-modality scenarios, we simulate a more realistic setting by considering both client-level and instance-level modality incompleteness in this study. Building on this realistic simulation, we propose ClusMFL, a novel MFL framework that leverages feature clustering for cross-institutional brain imaging analysis under modality incompleteness. Specifically, ClusMFL utilizes the FINCH algorithm to construct a pool of cluster centers for the feature embeddings of each modality-label pair, effectively capturing fine-grained data distributions. These cluster centers are then used for feature alignment within each modality through supervised contrastive learning, while also acting as proxies for missing modalities, allowing cross-modal knowledge transfer. Furthermore, ClusMFL employs a modality-aware aggregation strategy, further enhancing the model's performance in scenarios with severe modality incompleteness. We evaluate the proposed framework on the ADNI dataset, utilizing structural MRI and PET scans. Extensive experimental results demonstrate that ClusMFL achieves state-of-the-art performance compared to various baseline methods across varying levels of modality incompleteness, providing a scalable solution for cross-institutional brain imaging analysis.

eess.IV↗

arXiv2025

RADICE: Causal Graph Based Root Cause Analysis for System Performance Diagnostic

Root cause analysis is one of the most crucial operations in software reliability regarding system performance diagnostic. It aims to identify the root causes of system performance anomalies, allowing the resolution or the future prevention of issues that can cause millions of dollars in losses. Common existing approaches relying on data correlation or full domain expert knowledge are inaccurate or infeasible in most industrial cases, since correlation does not imply causation, and domain experts may not have full knowledge of complex and real-time systems. In this work, we define a novel causal domain knowledge model representing causal relations about the underlying system components to allow domain experts to contribute partial domain knowledge for root cause analysis. We then introduce RADICE, an algorithm that through the causal graph discovery, enhancement, refinement, and subtraction processes is able to output a root cause causal sub-graph showing the causal relations between the system components affected by the anomaly. We evaluated RADICE with simulated data and reported a real data use case, sharing the lessons we learned. The experiments show that RADICE provides better results than other baseline methods, including causal discovery algorithms and correlation based approaches for root cause analysis.

cs.SE↗

arXiv2024

TTT-Unet: Enhancing U-Net with Test-Time Training Layers for Biomedical Image Segmentation

Biomedical image segmentation is crucial for accurately diagnosing and analyzing various diseases. However, Convolutional Neural Networks (CNNs) and Transformers, the most commonly used architectures for this task, struggle to effectively capture long-range dependencies due to the inherent locality of CNNs and the computational complexity of Transformers. To address this limitation, we introduce TTT-Unet, a novel framework that integrates Test-Time Training (TTT) layers into the traditional U-Net architecture for biomedical image segmentation. TTT-Unet dynamically adjusts model parameters during the testing time, enhancing the model's ability to capture both local and long-range features. We evaluate TTT-Unet on multiple medical imaging datasets, including 3D abdominal organ segmentation in CT and MR images, instrument segmentation in endoscopy images, and cell segmentation in microscopy images. The results demonstrate that TTT-Unet consistently outperforms state-of-the-art CNN-based and Transformer-based segmentation models across all tasks. The code is available at https://github.com/rongzhou7/TTT-Unet.

eess.IV↗

arXiv2024

Normative Modeling for AD Diagnosis and Biomarker Identification

In this paper, we introduce a novel normative modeling approach that incorporates focal loss and adversarial autoencoders (FAAE) for Alzheimer's Disease (AD) diagnosis and biomarker identification. Our method is an end-to-end approach that embeds an adversarial focal loss discriminator within the autoencoder structure, specifically designed to effectively target and capture more complex and challenging cases. We first use the enhanced autoencoder to create a normative model based on data from healthy control (HC) individuals. We then apply this model to estimate total and regional neuroanatomical deviation in AD patients. Through extensive experiments on the OASIS-3 and ADNI datasets, our approach significantly outperforms previous state-of-the-art methods. This advancement not only streamlines the detection process but also provides a greater insight into the biomarker potential for AD. Our code can be found at \url{https://github.com/soz223/FAAE}.

eess.IV↗

arXiv2024

Monte Carlo Simulation of Angular Response of GRID Detectors for GRID Mission

The Gamma-Ray Integrated Detectors (GRID) are a space science mission that employs compact gamma-ray detectors mounted on NanoSats in low Earth orbit (LEO) to monitor the transient gamma-ray sky. Owing to the unpredictability of the time and location of gamma-ray bursts (GRBs), obtaining the photon responses of gamma-ray detectors at various incident angles is important for the scientific analysis of GRB data captured by GRID detectors. For this purpose, a dedicated Monte Carlo simulation framework has been developed for GRID detectors. By simulating each GRID detector and the NanoSat carrying it, the spectral energy response, detection efficiency, and other angular responses of each detector for photons with different incident angles and energies can be obtained within this framework. The accuracy of these simulations has been corroborated through on-ground calibration, and the derived angular responses have been successfully applied to the data analysis of recorded GRBs.

astro-ph.IM↗

arXiv2024

Independence of $\ell$ for Frobenius conjugacy classes attached to abelian varieties

Let $A$ be an abelian variety over a number field $\mathrm E\subset \mathbb C$ and let $\mathbf G$ denote the Mumford--Tate group of $A$. After replacing $\mathrm E$ by a finite extension, the action of the absolute Galois group $\mathrm{Gal}(\overline{\mathrm E}/\mathrm E)$ on the $\ell$-adic cohomology $\mathrm{H}^1_{\mathrm{\acute{e}t}}(A_{\overline{\mathrm E}},\mathbb Q_\ell)$ factors through $\mathbf G(\mathbb Q_\ell).$ We show that for $v$ an odd prime of $\mathrm E$ where $A$ has good reduction, the conjugacy class of Frobenius $\mathrm{Frob}_v$ in $\mathbf G(\mathbb Q_\ell)$ is independent of $\ell$. Along the way we prove that every point in the $μ$-ordinary locus of the special fiber of Shimura varieties has a special point lifting it.

math.NT↗

arXiv2024

Biomedical SAM 2: Segment Anything in Biomedical Images and Videos

Medical image segmentation and video object segmentation are essential for diagnosing and analyzing diseases by identifying and measuring biological structures. Recent advances in natural domain have been driven by foundation models like the Segment Anything Model 2 (SAM-2). To explore the performance of SAM-2 in biomedical applications, we designed three evaluation pipelines for single-frame 2D image segmentation, multi-frame 3D image segmentation and multi-frame video segmentation with varied prompt designs, revealing SAM-2's limitations in medical contexts. Consequently, we developed BioSAM-2, an enhanced foundation model optimized for biomedical data based on SAM-2. Our experiments show that BioSAM-2 not only surpasses the performance of existing state-of-the-art foundation models but also matches or even exceeds specialist models, demonstrating its efficacy and potential in the medical domain.

cs.CV↗

arXiv2024

BiomedGPT: A Generalist Vision-Language Foundation Model for Diverse Biomedical Tasks

Traditional biomedical artificial intelligence (AI) models, designed for specific tasks or modalities, often exhibit limited flexibility in real-world deployment and struggle to utilize holistic information. Generalist AI holds the potential to address these limitations due to its versatility in interpreting different data types and generating tailored outputs for diverse needs. However, existing biomedical generalist AI solutions are typically heavyweight and closed source to researchers, practitioners, and patients. Here, we propose BiomedGPT, the first open-source and lightweight vision-language foundation model, designed as a generalist capable of performing various biomedical tasks. BiomedGPT achieved state-of-the-art results in 16 out of 25 experiments while maintaining a computing-friendly model scale. We also conducted human evaluations to assess the capabilities of BiomedGPT in radiology visual question answering, report generation, and summarization. BiomedGPT exhibits robust prediction ability with a low error rate of 3.8% in question answering, satisfactory performance with an error rate of 8.3% in writing complex radiology reports, and competitive summarization ability with a nearly equivalent preference score to human experts. Our method demonstrates that effective training with diverse data can lead to more practical biomedical AI for improving diagnosis and workflow efficiency.

cs.CL↗

arXiv2024

ViT-1.58b: Mobile Vision Transformers in the 1-bit Era

Vision Transformers (ViTs) have achieved remarkable performance in various image classification tasks by leveraging the attention mechanism to process image patches as tokens. However, the high computational and memory demands of ViTs pose significant challenges for deployment in resource-constrained environments. This paper introduces ViT-1.58b, a novel 1.58-bit quantized ViT model designed to drastically reduce memory and computational overhead while preserving competitive performance. ViT-1.58b employs ternary quantization, which refines the balance between efficiency and accuracy by constraining weights to {-1, 0, 1} and quantizing activations to 8-bit precision. Our approach ensures efficient scaling in terms of both memory and computation. Experiments on CIFAR-10 and ImageNet-1k demonstrate that ViT-1.58b maintains comparable accuracy to full-precision Vit, with significant reductions in memory usage and computational costs. This paper highlights the potential of extreme quantization techniques in developing sustainable AI solutions and contributes to the broader discourse on efficient model deployment in practical applications. Our code and weights are available at https://github.com/DLYuanGod/ViT-1.58b.

cs.CV↗

arXiv2024

Towards an Extensible Model-Based Digital Twin Framework for Space Launch Vehicles

The concept of Digital Twin (DT) is increasingly applied to systems on different levels of abstraction across domains, to support monitoring, analysis, diagnosis, decision making and automated control. Whilst the interest in applying DT is growing, the definition of DT is unclear, neither is there a clear pathway to develop DT to fully realise its capacities. In this paper, we revise the concept of DT and its categorisation. We propose a DT maturity matrix, based on which we propose a model-based DT development methodology. We also discuss how model-based tools can be used to support the methodology and present our own supporting tool. We report our preliminary findings with a discussion on a case study, in which we use our proposed methodology and our supporting tool to develop an extensible DT platform for the assurance of Electrical and Electronics systems of space launch vehicles.

cs.SE↗

arXiv2024

Unleashing the Power of Multi-Task Learning: A Comprehensive Survey Spanning Traditional, Deep, and Pretrained Foundation Model Eras

MTL is a learning paradigm that effectively leverages both task-specific and shared information to address multiple related tasks simultaneously. In contrast to STL, MTL offers a suite of benefits that enhance both the training process and the inference efficiency. MTL's key advantages encompass streamlined model architecture, performance enhancement, and cross-domain generalizability. Over the past twenty years, MTL has become widely recognized as a flexible and effective approach in various fields, including CV, NLP, recommendation systems, disease prognosis and diagnosis, and robotics. This survey provides a comprehensive overview of the evolution of MTL, encompassing the technical aspects of cutting-edge methods from traditional approaches to deep learning and the latest trend of pretrained foundation models. Our survey methodically categorizes MTL techniques into five key areas: regularization, relationship learning, feature propagation, optimization, and pre-training. This categorization not only chronologically outlines the development of MTL but also dives into various specialized strategies within each category. Furthermore, the survey reveals how the MTL evolves from handling a fixed set of tasks to embracing a more flexible approach free from task or modality constraints. It explores the concepts of task-promptable and -agnostic training, along with the capacity for ZSL, which unleashes the untapped potential of this historically coveted learning paradigm. Overall, we hope this survey provides the research community with a comprehensive overview of the advancements in MTL from its inception in 1997 to the present in 2023. We address present challenges and look ahead to future possibilities, shedding light on the opportunities and potential avenues for MTL research in a broad manner. This project is publicly available at https://github.com/junfish/Awesome-Multitask-Learning.

cs.LG↗

arXiv2024

On the smooth locus of affine Schubert varieties

We give a simple and uniform proof of a conjecture of Haines-Richarz characterizing the smooth locus of Schubert varieties in twisted affine Grassmannians. Our method is elementary and avoids any representation theoretic techniques, instead relying on a combinatorial analysis of tangent spaces of Schubert varieties.

math.AG↗

arXiv2024

Geometric inequalities and their stabilities for modified quermassintegrals in hyperbolic space

In this paper, we first consider the curve case of Hu-Li-Wei's flow for shifted principal curvatures of h-convex hypersurfaces in $\mathbb{H}^{n+1}$ proposed in [10]. We prove that if the initial closed curve is smooth and strictly h-convex, then the solution exists for all time and preserves strict h-convexity along the flow. Moreover, the evolving curve converges smoothly and exponentially to a geodesic circle centered at the origin. The key ingredient in our proof is the Heintze-Karcher type inequality for h-convex curves proved recently in [14]. As an application, we then provide a new proof of geometric inequalities involving weighted curvature integrals and modified quermassintegrals for h-convex curves in $\mathbb{H}^2$. We finally discuss the stability of these inequalities as well as Alexandrov-Fenchel type inequalities for modified quermassintegrals for strictly h-convex domains in $\mathbb{H}^{n+1}$.

math.DG↗

arXiv2024

Gauss curvature type flow and Alexandrov-Fenchel inequalities in the hyperbolic space

We consider the Gauss curvature type flow for uniformly convex hypersurfaces in the hyperbolic space $\mathbb{H}^{n+1}\ (n\geqslant 2)$. We prove that if the initial closed hypersurface is smooth and uniformly convex, then the smooth solution exists for all positive time and converges smoothly and exponentially to a geodesic sphere centered at the origin. The key step is to prove the upper bound of Gauss curvature and that uniform convexity preserves along the flow. As an application, we provide a new proof for an Alexandrov-Fenchel inequality comparing $(n-2)$th quermassintegral and volume of convex domains in $\mathbb{H}^{n+1}\ (n\geqslant 2)$.

math.DG↗