检索结果-内蒙古大学图书馆

51st International Colloquium on Automata, Languages, and Programming, ICALP 2024

作者： Czumaj, Artur Gao, Guichen Jiang, Shaofeng H.-C. Krauthgamer, Robert Veselý, Pavel Department of Computer Science University of Warwick Coventry United Kingdom School of Computer Science Peking University Beijing China Department of Computer Science and Applied Mathematics Weizmann Institute of Science Rehovot Israel Computer Science Institute of Charles University Prague Czech Republic

ISBN: (纸本)9783959773225

We design new parallel algorithms for clustering in high-dimensional Euclidean spaces. These algorithms run in the Massively Parallel Computation (MPC) model, and are fully scalable, meaning that the local memory in each machine may be nσ for arbitrarily small fixed σ > 0. Importantly, the local memory may be substantially smaller than the number of clusters k, yet all our algorithms are fast, i.e., run in O(1) rounds. We first devise a fast MPC algorithm for O(1)-approximation of uniform Facility Location. This is the first fully-scalable MPC algorithm that achieves O(1)-approximation for any clustering problem in general geometric setting;previous algorithms only provide poly(log n)-approximation or apply to restricted inputs, like low dimension or small number of clusters k;e.g. [Bhaskara and Wijewardena, ICML’18;Cohen-Addad et al., NeurIPS’21;Cohen-Addad et al., ICML’22]. We then build on this Facility Location result and devise a fast MPC algorithm that achieves O(1)-bicriteria approximation for k-Median and for k-Means, namely, it computes (1 + Ε)k clusters of cost within O(1/Ε2)-factor of the optimum for k clusters. A primary technical tool that we introduce, and may be of independent interest, is a new MPC primitive for geometric aggregation, namely, computing for every data point a statistic of its approximate neighborhood, for statistics like range counting and nearest-neighbor search. Our implementation of this primitive works in high dimension, and is based on consistent hashing (aka sparse partition), a technique that was recently used for streaming algorithms [Czumaj et al., FOCS’22]. © Artur Czumaj, Guichen Gao, Shaofeng H.-C. Jiang, Robert Krauthgamer, and Pavel Veselý.

关键词： Nearest neighbor search

来源：评论

学校读者我要写书评

暂无评论

Nonlinearity in Large-Cycle Cellular Automata

引用

Complex Systems 2025年第1期34卷 129-160页

作者： Mukherjee, Sukanya Adak, Sumit Department of Computer Science and Engineering Institute of Engineering and Management Kolkata University of Engineering and Management West Bengal Kolkata700091 India Department of Applied Mathematics and Computer Science DTU Compute Technical University of Denmark Kgs. Lyngby2800 Denmark

The class of maximal-length cellular automata (CAs) has gained significant attention over the last few years due to the fact that it can generate cycles with the longest possible lengths. For every l of the form l = 2n-1 where n ∈ , a cycle of length l can be generated by a cellular automaton (CA) of size ⌊log2 l⌋ + 1;this maximal-length CA is in fact the optimal (minimum-sized) CA to generate a cycle of length l. However, maximal-length CAs are, in general, linear and have some drawbacks. Moreover, it is nontrivial to address this problem for any l ∈ , that is, removing the constraint that l is of the form 2n-1. In this paper we therefore employ nonlinearity in the rules and aim to generate the near-optimal CA for any given cycle length l, meaning that the generated CAs will be of size n = (⌊log2 l⌋ + 1) + d, where the displacement d is as small as possible. We use a generic operator to concatenate a collection of component CAs of sizes n1, n2, …, nk by employing an evolutionary strategy. Here, n = n1 + n2 + ⋯ + nk and l ≥ lcm(1, 2, …, k ), where i (1 ≤ i ≤ k) is the maximum cycle length of a CA of size ni . An influencing factor in the design of component CAs is to achieve the near-optimal CA for a given cycle length. Here we consider nonlinear reversible large-cycle CAs as component CAs. We also measure the effectiveness of nonlinearity in large-cycle generation. © 2025, Complex Systems Publications, Inc. All rights reserved.

关键词： CAs cellular automata evolution large-cycle CA neighborhood dependent nonlinearity reversible CAs

来源：评论

学校读者我要写书评

暂无评论

Learning to select the recombination operator for derivative-free optimization

引用

science China mathematics 2024年第6期67卷 1457-1480页

作者： Haotian Zhang Jianyong Sun Thomas Back Zongben Xu School of Mathematics and Statistics Xi’an Jiaotong UniversityXi’an710049China Leiden Institute of Advanced Computer Science Leiden UniversityLeiden2333 CANetherlands

Extensive studies on selecting recombination operators adaptively,namely,adaptive operator selection(AOS),during the search process of an evolutionary algorithm(EA),have shown that AOS is promising for improving EA's performance.A variety of heuristic mechanisms for AOS have been proposed in recent decades,which usually contain two main components:the feature extraction and the policy *** feature extraction refers to as extracting relevant features from the information collected during the search *** policy setting means to set a strategy(or policy)on how to select an operator from a pool of operators based on the extracted *** components are designed by hand in existing studies,which may not be efficient for adapting optimization *** this paper,a generalized framework is proposed for learning the components of AOS for one of the main streams of EAs,namely,differential evolution(DE).In the framework,the feature extraction is parameterized as a deep neural network(DNN),while a Dirichlet distribution is considered to be the policy.A reinforcement learning method,named policy gradient,is used to train the *** case studies,the proposed framework is applied to two DEs including the classic DE and a recently-proposed DE,which result in two new algorithms named PG-DE and PG-MPEDE,*** on the Congress of Evolutionary Computation(CEC)2018 test suite show that the proposed new algorithms perform significantly better than their ***,we prove theoretically that the considered classic methods are the special cases of the proposed framework.

关键词： evolutionary algorithm differential evolution adaptive operator selection reinforcement learning deep learning

来源：评论

学校读者我要写书评

暂无评论

An Anchor-Aware Graph Autoencoder Fused with Gini Index Model for Link Prediction

引用

SN computer science 2024年第6期5卷 1-16页

作者： Kumar, Shambhu Bisht, Dinesh Jain, Arti Computer Science & amp Engineering and IT Jaypee Institute of Information Technology Department of Mathematics Jaypee Institute of Information Technology

Link prediction stands as a crucial network challenge, garnering attention over the past decade, with its significance heightened by the escalating volume of network data. In response to the pressing need for swift research focus, this study introduces an innovative approach—the Anchor-aware Graph Autoencoder integrated with the Gini Index (AGA-GI)—aimed at gathering data on the global placements of link nodes within the link prediction framework. The proposed methodology encompasses three key components: anchor points, node-to-anchor paths, and node embedding. Anchor points within the network are identified by leveraging the graph structure as an input. The determination of anchor positions involves computing the Gini indexes (GI) of nodes, leading to the generation of a candidate set of anchors. Typically, these anchor points are distributed across the network structure, facilitating substantial informational exchanges with other nodes. The location-based similarity approach computes the paths between anchor points and nodes. It identifies the shortest path, creating a node path information function that incorporates feature details and location similarity. The ultimate embedding representation of the node is then formed by amalgamating attributes, global location data, and neighbourhood structure through an auto-encoder learning methodology. The Residual Capsule Network (RCN) model acquires these node embeddings as input to learn the feature representation of nodes and transforms the link prediction problem into a classification task. The suggested (AGA-GI) model undergoes comparison with various existing models in the realm of link prediction. These models include Attributes for Link Prediction (SEAL), Embeddings, Subgraphs, Dual-Encoder graph embedding with Alignment (DEAL), Embeddings and Spectral Clustering (SC), Deep Walk (DW), Graph Auto-encoder (GAE), Variational Graph Autoencoders (VGAE), Graph Attention Network (GAT), and Graph Conversion Capsule Link (G

关键词： Capsule network Ideal Anchor Point Link prediction Path Estimation Portion Path information function Residual module Shortest path distance

来源：评论

学校读者我要写书评

暂无评论

Security and Privacy in Machine Learning for Health Systems: Strategies and Challenges

引用

Yearbook of Medical Informatics 2023年第1期32卷 269-281页

作者： de Aguiar, Erikson J. Traina, Caetano Traina, Agma J.M. Institute of Mathematics and Computer Science University of São Paulo Brazil

OBJECTIVES: Machine learning (ML) is a powerful asset to support physicians in decision-making procedures, providing timely answers. However, ML for health systems can suffer from security attacks and privacy violations. This paper investigates studies of security and privacy in ML for health. METHODS: We examine attacks, defenses, and privacy-preserving strategies, discussing their challenges. We conducted the following research protocol: starting a manual search, defining the search string, removing duplicated papers, filtering papers by title and abstract, then their full texts, and analyzing their contributions, including strategies and challenges. Finally, we collected and discussed 40 papers on attacks, defense, and privacy. RESULTS: Our findings identified the most employed strategies for each domain. We found trends in attacks, including universal adversarial perturbation (UAPs), generative adversarial network (GAN)-based attacks, and DeepFakes to generate malicious examples. Trends in defense are adversarial training, GAN-based strategies, and out-of-distribution (OOD) to identify and mitigate adversarial examples (AE). We found privacy-preserving strategies such as federated learning (FL), differential privacy, and combinations of strategies to enhance the FL. Challenges in privacy comprehend the development of attacks that bypass fine-tuning, defenses to calibrate models to improve their robustness, and privacy methods to enhance the FL strategy. CONCLUSIONS: In conclusion, it is critical to explore security and privacy in ML for health, because it has grown risks and open vulnerabilities. Our study presents strategies and challenges to guide research to investigate issues about security and privacy in ML applied to health systems. IMIA and Thieme. This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial License, permitting copying and reproduction so long as the original work is gi

关键词： Adversarial attacks machine learning medical images privacy

来源：评论

学校读者我要写书评

暂无评论

Notes on Linear Gossip Algorithms:Potentials and Challenges

引用

Journal of Systems science & Complexity 2025年第1期38卷 98-128页

作者： GERENCSER Balazs GERENCSER Laszlo HUN-REN Alfred Renyi Institute of Mathematics Budapest 1052Hungary and Eotvos Lorand UniversityDepartment of Probability and StatisticsBudapest 1518Hungary HUN-REN SZTAKI Institute for Computer Science and ControlBudapest 1111Hungary

The authors consider the problem of reaching consensus over a communication network via asynchronous interaction between pairs of agents.A well-known method is the linear gossip algorithm due to Tsitsiklis(1984).Extension of this,allowing the selection of a strictly stationary sequence of communicating pairs,was given in Picci and Taylor(2013).Extension of the linear gossip algorithm to directed communication networks,retaining the linear dynamics,was proposed by Cai and Ishii(2012),later extended by Silvestre,et al.(2018).A definite novelty of these algorithms is that L2-convergence with exponential rate can be *** authors attend the above issues,extending the result of Picci and Taylor(2013)motivated by features of algorithms for directed *** authors present and discuss the algorithm of Silvestre,et al.(2018),together with systematic simulation results based on 5M randomly chosen parameter *** core of the proposed mathematical technology is a set of simple observations,presented with a tutorial aspect,by which the authors can conveniently establish various results on the almost sure convergence of products of strictly stationary sequences of matrices to a rank-1 matrix.

关键词： Consensus gossip algorithms Lyapunov-exponents spectral gap

来源：评论

学校读者我要写书评

暂无评论

Dual Layer Cogni - Insight Deep-Mood Encoder: A Two- Tiered Approach for Depression Detection

Dual Layer Cogni - Insight Deep-Mood Encoder: A Two- Tiered ...

引用

2024 International Russian Smart Industry Conference, SmartIndustryCon 2024

作者： Firoz, Neda Berestneva, Olga Aksyonov, Sergey V. Institute of Applied Mathematics and Computer Science Tomsk State University Tomsk Russia

ISBN: (纸本)9798350395044

Depression (MDD) affects approximately 5% of adults globally, contributing to productivity loss and public health concerns. With 280 million people impacted, the risk of suicide and self-harm underscores its severity. Current treatments are inadequate, and the complex, heterogeneous nature of depression necessitates automated solutions for timely assessment and varied treatments. This study explores the transformative role of Artificial Intelligence in detecting depressive symptoms, leveraging NLP, and multimodal data analysis. The research introduces the 'Dual Layer Cogni-Insight Deep-Mood Encoder (DLCDME)' model, incorporating two-tiered feature encodings. Four pre-trained language models (BERT, MentalBERT, MentalRoBERTa, PsychBERT, and ClinicalBERT) are experimented and fused with linguistic features for comprehensive representation. A transformer encoder, multi-head attention, and LSTM fine-tuning enhances contextual text feature embeddings, demonstrating the efficacy of ClinicalBERT with transformer encoders for cost-effective large corpus handling. Contributions include a novel feature extraction approach, a thorough model analysis, and the superior performance of DLCDME over contemporary works, holding promise for clinical diagnosis. Our findings reveal that a) the application of ClinicalBERT over DLCDME significantly enhances the performance of depression detection, and b) employing dual feature encodings on text data with transformer attention for predicting depression yields an improved MAE of 3.40 and RMSE of 4.38. The model achieved a test accuracy of 95%, with F1 scores for precision, and recall recorded at 0.95, 0.96, 0.95 respectively. © 2024 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

A CNN Model Based Approach for Disease Detection in Mango Plant Leaves 12th

A CNN Model Based Approach for Disease Detection in Mango Pl...

引用

12th International Conference on Soft Computing for Problem Solving, SocProS 2023

作者： Singh, Pankaj Pratap Kumar, Devanshu Srivastava, Aakriti Basumatary, Madhusmita Prasad, Shitala Department of Computer Science and Engineering Central Institute of Technology Kokrajhar Assam India School of Mathematics and Computer Science Indian Institute of Technology Goa Goa India

ISBN: (纸本)9789819732913

In this study, we propose a machine learning (ML) based method for the early detection of plant leaf diseases. Plant diseases are a major concern in agriculture, impacting crop yield, and food security. Early and accurate identification of these diseases is vital for effective disease management. Our approach employs deep learning algorithms and a dataset of annotated images containing both healthy and diseased plant leaves. Through training a deep neural network, we enable automated disease detection by extracting relevant features from leaf images and classifying them as healthy or diseased. By reducing reliance on human expertise, our approach enables timely detection, facilitating prompt implementation of disease management strategies. This ML-based method has the potential to revolutionize the field of plant pathology, offering valuable insights for the development of precision agriculture techniques aimed at ensuring sustainable crop production. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

ON MODELLING OF THREE DIMENSIONAL FLOW IN EXTRACTION OF BIOLOGICALLY ACTIVE SUBSTANCES FROM PLANTS 22

ON MODELLING OF THREE DIMENSIONAL FLOW IN EXTRACTION OF BIOL...

引用

22nd International Scientific Conference Engineering for Rural Development, ERD 2023

作者： Strautins, Uldis Marinaki, Maksims Institute of Mathematics and Computer Science University of Latvia Latvia

The authors consider the process of extraction from heterogeneous plant material;a novel model offering 3D spatial resolution is proposed and simulation possibilities by applying several modern numerical methods for three dimensional problems are discussed. These types of problems naturally arise when one wishes to mathematically describe the processes of the extraction with solvents such as supercritical carbon dioxide (scCO2). The results of this process meet the demand of the functional foods, nutraceutical, pharmaceutical and cosmetic industries on bioactive compounds. The models used in typical simulations by the industry offer no spatial resolution of the extraction process, however, when novel designs of extractors are developed, the three dimensional flow of the solvent and its interaction with the plant material should be considered. We propose a coupled system of equations consisting of fluid dynamics equations describing the flow of the solvent and reactionadvection-diffusion equations for the solute, plus equations describing the remaining concentration of solute in the biomass. The exchange of the active material between the solid and fluid is modelled by the Langmuir law. The initial distribution of solute in the plant material is not assumed to be homogeneous, nor are the extraction parameters, i.e. heterogeneous mixture of plant materials can be covered by this model. The equations are discretized by using finite volume techniques and currently implemented in Octave/Matlab environment. As a result, the temporal evolution of the three dimensional distribution of the solute is obtained along with the velocity field of the solvent, allowing detailed studies of the extraction process in a virtual design. The model is modular, allowing integrating tailor-made laws to describe various plant materials. The authors present simulation results and analyse the advantages and disadvantages of the proposed methods and approaches. © 2023 Latvia University of Life

关键词： Numerical methods

来源：评论

学校读者我要写书评

暂无评论

Designing Exploration Contracts 42

Designing Exploration Contracts

引用

42nd International Symposium on Theoretical Aspects of computer science, STACS 2025

作者： Hoefer, Martin Schecker, Conrad Schewior, Kevin Department of Computer Science RWTH Aachen University Germany Institute for Computer Science Goethe University Frankfurt Germany Department of Mathematics and Computer Science University of Cologne Germany University of Southern Denmark Odense Denmark

ISBN: (纸本)9783959773652

We study a natural application of contract design in the context of sequential exploration problems. In our principal-agent setting, a search task is delegated to an agent. The agent performs a sequential exploration of n boxes, suffers the exploration cost for each inspected box, and selects the content (called the prize) of one inspected box as outcome. Agent and principal obtain an individual value based on the selected prize. To influence the search, the principal a-priori designs a contract with a non-negative payment to the agent for each potential prize. The goal of the principal is to maximize her expected reward, i.e., value minus payment. Interestingly, this natural contract scenario shares close relations with the Pandora’s Box problem. We show how to compute optimal contracts for the principal in several scenarios. A popular and important subclass is that of linear contracts, and we show how to compute optimal linear contracts in polynomial time. For general contracts, we obtain optimal contracts under the standard assumption that the agent suffers cost but obtains value only from the transfers by the principal. More generally, for general contracts with non-zero agent values for outcomes we show how to compute an optimal contract in two cases: (1) when each box has only one prize with non-zero value for principal and agent, (2) for i.i.d. boxes with a single prize with positive value for the principal. © Martin Hoefer, Conrad Schecker, and Kevin Schewior.

关键词： Contract Design Exploration Pandora’s Box Problem

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：