检索结果-内蒙古大学图书馆

An effective method for quantification,visualization,and analysis of 3D cell shape during early embryogenesis

学校读者我要写书评

暂无评论

Quantitative Biology 2025年第1期13卷 113-126页

作者： Zelin Li Zhaoke Huang Jianfeng Cao Guoye Guan Zhongying Zhao Hong Yan Department of Electrical Engineering City University of Hong KongHong KongChina Centre for Intelligent Multidimensional Data Analysis Limited Hong KongChina Center for Quantitative Biology Peking UniversityBeijingBeijingChina Department of Systems Biology Harvard Medical SchoolBostonMassachusettsUSA Department of Data Science Dana-Farber Cancer InstituteBostonMassachusettsUSA Department of Biology Hong Kong Baptist UniversityHong KongChina State Key Laboratory of Environmental and Biological Analysis Hong Kong Baptist UniversityHong KongChina

Embryogenesis is the most basic process in developmental *** and simply quantifying cell shape is challenging for the complex and dynamic 3D embryonic *** descriptors such as volume,surface area,and mean curvature often fall short,providing only a global view and lacking in local detail and reconstruction *** this,we introduce an effective integrated method,3D Cell Shape Quantification(3DCSQ),for transforming digitized 3D cell shapes into analytical feature vectors,named eigengrid(proposed grid descriptor like eigen value),eigenharmonic,and *** uniquely combine spherical grids,spherical harmonics,and principal component analysis for cell shape *** demonstrate 3DCSQ’s effectiveness in recognizing cellular morphological phenotypes and clustering *** to Caenorhabditis elegans embryos of 29 living embryos from 4-to 350-cell stages,3DCSQ identifies and quantifies biologically reproducible cellular patterns including distinct skin cell *** also provide automatically cell shape lineaging analysis *** method not only systematizes cell shape description and evaluation but also monitors cell differentiation through shape changes,presenting an advancement in biological imaging and analysis.

关键词： Caenorhabditis elegans(C.elegans) cell shape quantification eigen features(eigengrid eigenharmonic&eigenspectrum) lineage analysis morphological reproducibility spherical harmonics(SPHARM)

ELFATT: Efficient Linear Fast Attention for Vision Transformers

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wu, Chong Che, Maolin Xu, Renjie Ran, Zhuoheng Yan, Hong Department of Electrical Engineering Centre for Intelligent Multidimensional Data Analysis City University of Hong Kong Kowloon Hong Kong School of Mathematics and Statistics Guizhou University Guiyang550025 China

The attention mechanism is the key to the success of transformers in different machine learning tasks. However, the quadratic complexity with respect to the sequence length of the vanilla softmax-based attention mechanism becomes the major bottleneck for the application of long sequence tasks, such as vision tasks. Although various efficient linear attention mechanisms have been proposed, they need to sacrifice performance to achieve high efficiency. What’s more, memory-efficient methods, such as FlashAttention-1-3, still have quadratic computation complexity which can be further improved. In this paper, we propose a novel efficient linear fast attention (ELFATT) mechanism to achieve low memory input/output operations, linear computational complexity, and high performance at the same time. ELFATT offers 4-7x speedups over the vanilla softmax-based attention mechanism in high-resolution vision tasks without losing performance. ELFATT is FlashAttention friendly. Using FlashAttention-2 acceleration, ELFATT still offers 2-3x speedups over the vanilla softmax-based attention mechanism on high-resolution vision tasks without losing performance. Even on edge GPUs, ELFATT still offers 1.6x to 2.0x speedups compared to state-of-the-art attention mechanisms in various power modes from 5W to 60W. Furthermore, ELFATT can be used to enhance and accelerate diffusion tasks directly without training. Copyright © 2025, The Authors. All rights reserved.

关键词： Machine vision

Causal Inference based Transfer Learning with LLMs: An Efficient Framework for Industrial RUL Prediction

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Chen, Yan Liu, Cheng Department of Systems Engineering City University of Hong Kong Hong Kong Centre for Intelligent Multidimensional Data Analysis City University of Hong Kong Hong Kong

Accurate prediction of Remaining Useful Life (RUL) for complex machinery is critical for industrial prognostics but challenged by high-dimensional, noisy sensor data. We propose the Causal-Informed data Pruning Framework (CIDPF). It pioneers causal inference for industrial data pruning, identifying sensor signals with robust causal relationships to RUL via PCMCI-based stability analysis, while a Gaussian Mixture Model (GMM) screens anomalies. By training on only 10% of data after pruning, CIDPF fine-tunes pre-trained LLMs with parameter-efficient strategies, reducing training time by 90% compared to traditional approach. Experiments on N-CMAPSS show CIDPF achieves 26% lower RMSE than existing methods and 25% improvement over full-data baselines, demonstrating superior accuracy with drastically reduced computational costs. The framework’s adaptability to multi-condition scenarios underscores its practicality for industrial deployment. Copyright © 2025, The Authors. All rights reserved.

关键词： Transfer learning

Few-shot crack image classification using clip based on bayesian optimization

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhang, Yingchao Liu, Cheng Department of Systems Engineering City University of Hong Kong Hong Kong Centre for Intelligent Multidimensional Data Analysis City University of Hong Kong Hong Kong

This study proposes a novel few-shot crack image classification model based on CLIP and Bayesian optimization. By combining multimodal information and Bayesian approach, the model achieves efficient classification of crack images in a small number of training samples. The CLIP model employs its robust feature extraction capabilities to facilitate precise classification with a limited number of samples. In contrast, Bayesian optimisation enhances the robustness and generalization of the model, while reducing the reliance on extensive labelled data. The results demonstrate that the model exhibits robust performance across a diverse range of dataset scales, particularly in the context of small sample sets. The study validates the potential of the method in civil engineering crack *** Codes 68T07 © 2025, CC BY-NC-ND.

关键词： Image classification

Sequential Multi-objective Multi-agent Reinforcement Learning Approach for Predictive Maintenance

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Chen, Yan Liu, Cheng Department of Systems Engineering City University of Hong Kong Hong Kong Centre for Intelligent Multidimensional Data Analysis City University of Hong Kong Hong Kong

Existing predictive maintenance (PdM) methods typically focus solely on whether to replace system components without considering the costs incurred by inspection. However, a well-considered approach should be able to minimize Remaining Useful Life (RUL) at engine replacement while maximizing inspection interval. To achieve this, multi-agent reinforcement learning (MARL) can be introduced. However, due to the sequential and mutually constraining nature of these 2 objectives, conventional MARL is not applicable. Therefore, this paper introduces a novel framework and develops a Sequential Multi-objective Multi-agent Proximal Policy Optimization (SMOMA-PPO) algorithm. Furthermore, to provide comprehensive and effective degradation information to RL agents, we also employed Gated Recurrent Unit, quantile regression, and probability distribution fitting to develop a GRU-based RUL Prediction (GRP) model. Experiments demonstrate that the GRP method significantly improves the accuracy of RUL predictions in the later stages of system operation compared to existing methods. When incorporating its output into SMOMA-PPO, we achieve at least a 15% reduction in average RUL without unscheduled replacements (UR), nearly a 10% increase in inspection interval, and an overall decrease in maintenance costs. Importantly, our approach offers a new perspective for addressing multi-objective maintenance planning with sequential constraints, effectively enhancing system reliability and reducing maintenance expenses. Copyright © 2025, The Authors. All rights reserved.

关键词： Predictive maintenance

TOWARDS SCALABLE TOPOLOGICAL REGULARIZERS

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wong, Hiu-Tung Lee, Darrick Yan, Hong Centre for Intelligent Multidimensional Data Analysis Science and Technology Park Hong Kong School of Mathematics University of Edinburgh United Kingdom Department of Electrical Engineering City University of Hong Kong Kowloon Hong Kong

Latent space matching, which consists of matching distributions of features in latent space, is a crucial component for tasks such as adversarial attacks and defenses, domain adaptation, and generative modelling. Metrics for probability measures, such as Wasserstein and maximum mean discrepancy, are commonly used to quantify the differences between such distributions. However, these are often costly to compute, or do not appropriately take the geometric and topological features of the distributions into consideration. Persistent homology is a tool from topological data analysis which quantifies the multi-scale topological structure of point clouds, and has recently been used as a topological regularizer in learning tasks. However, computation costs preclude larger scale computations, and discontinuities in the gradient lead to unstable training behavior such as in adversarial tasks. We propose the use of principal persistence measures, based on computing the persistent homology of a large number of small subsamples, as a topological regularizer. We provide a parallelized GPU implementation of this regularizer, and prove that gradients are continuous for smooth densities. Furthermore, we demonstrate the efficacy of this regularizer on shape matching, image generation, and semi-supervised learning tasks, opening the door towards a scalable regularizer for topological features. © 2025, CC BY.

关键词： Probability distributions

LU Decomposition and Generalized Autoone-Takagi Decomposition of Dual Matrices and their Applications

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Xu, Renjie Wei, Yimin Yan, Hong Center for Intelligent Multidimensional Data Analysis Hong Kong Science Park Shatin Hong Kong School of Mathematical Sciences Shanghai Key Laboratory of Contemporary Applied Mathematics Fudan University Shanghai200433 China Department of Electrical Engineering Center for Intelligent Multidimensional Data Analysis City University of Hong Kong Hong Kong

This paper uses matrix transformations to provide the Autoone-Takagi decomposition of dual complex symmetric matrices and extends it to dual quaternion η-Hermitian matrices. The LU decomposition of dual matrices is given using the general solution of the Sylvester equation, and its equivalence to the existence of rank-k decomposition and dual Moore-Penrose generalized inverse (DMPGI) is proved. Similar methods are then used to provide the Cholesky decomposition of dual real symmetric positive definite matrices. Both of our decompositions are driven by applications in numerical linear *** Codes 15A23, 15B33, 65F55 Copyright © 2025, The Authors. All rights reserved.

关键词：

Volume Tells: Dual Cycle-Consistent Diffusion for 3D Fluorescence Microscopy De-noising and Super-Resolution

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Li, Zelin Wang, Chenwei Huang, Zhaoke Ma, Yiming Zhao, Cunming Zhao, Zhongying Yan, Hong Department of Electrical Engineering City University of Hong Kong Hong Kong Centre for Intelligent Multidimensional Data Analysis Hong Kong Science Park Hong Kong Department of Biology Hong Kong Baptist University Hong Kong

3D fluorescence microscopy is essential for understanding fundamental life processes through long-term live-cell imaging. However, due to inherent issues in imaging principles, it faces significant challenges including spatially varying noise and anisotropic resolution, where the axial resolution lags behind the lateral resolution up to 4.5 times. Meanwhile, laser power is kept low to maintain cell viability, leading to inaccessible low-noise and high-resolution paired ground truth (GT). To tackle these limitations, a dual Cycle-consistent Diffusion is proposed to effectively mine intra-volume imaging priors within 3D cell volumes in an unsupervised manner, i.e., Volume Tells (VTCD), achieving de-noising and super-resolution (SR) simultaneously. Specifically, a spatially iso-distributed denoiser is designed to exploit the noise distribution consistency between adjacent low-noise and high-noise regions within the 3D cell volume, suppressing the spatially varying noise. Then, in light of the structural consistency of the cell volume, a cross-plane global-propagation SR module propagates high-resolution details from the XY plane into adjacent regions in the XZ and YZ planes, progressively enhancing resolution across the entire 3D cell volume. Experimental results on 10 in vivo cellular dataset demonstrate high improvements in both denoising and super-resolution, with axial resolution enhanced from ∼ 430 nm to ∼ 90 nm. © 2025, CC BY.

关键词： Fluorescence microscopy

Gradient neural network models for approximate Tucker decomposition of time-dependent tensors

学校读者我要写书评

暂无评论

Neurocomputing 2025年 640卷

作者： Che, Maolin Wei, Yimin Yan, Hong State Key Laboratory of Public Big Data and School of Mathematics and Statistics Guizhou University Guiyang550025 China Center for Intelligent Multidimensional Data Analysis Hong Science Park Shatin Hong Kong School of Mathematical Sciences and Key Laboratory of Mathematics for Nonlinear Sciences Fudan University Shanghai200433 China Department of Electrical Engineering and Center for Intelligent Multidimensional Data Analysis City University of Hong Kong 83 Tat Chee Avenue Kowloon Hong Kong

In this paper, we introduce a gradient neural network model for approximating the Tucker decomposition of time-dependent tensors with a specified multilinear rank bound. The behavior of this model is governed by a system of ordinary differential equations. First, based on the Karush–Kuhn–Tucker (KKT) conditions of the optimization problem for the approximate Tucker decomposition of a constant tensor, a gradient neural network model is proposed, demonstrating local stability according to Lyapunov stability theory. Next, we design a gradient neural network model specifically tailored for time-dependent tensors. Finally, using the Taylor-type difference rules, we discretize the proposed neural network models, ensuring zero stability for the implementation of digital hardware. We illustrate the effectiveness and characteristics of our approach through numerical examples. © 2025 Elsevier B.V.

关键词： Lyapunov methods