检索结果-内蒙古大学图书馆

作者： Iong, Daniel University of Michigan

学位级别：Ph.D., Doctor of Philosophy

Probabilistic models are able to encode our beliefs about underlying data generating mechanisms through distributional assumptions about model parameters and latent, or hidden, variables. Probabilistic models have had a huge impact on a wide range of fields such as epidemiology, biology, astronomy, psychology and much more. Due to their ability to encode complex structures, performing inference with these models can be challenging. This thesis proposes several inference algorithms for probabilistic models that are used to analyze data in epidemiology and space weather forecasting. In the second chapter, we propose a Monte-Carlo Expectation-Maximization (EM) algorithm for a novel mixture model used to model heterogeneous causal mechanisms in Mendelian Randomization (MR). We also derive approximate confidence intervals for uncertainty quantification, and adopt a modified Bayesian information criterion (BIC) for model selection. Our proposed model clusters the effects in a linear error-in-variables regression model using a Gaussian mixture model. We verify the efficacy of the Monte-Carlo EM algorithm, approximate confidence intervals, and model selection criterion through extensive simulation studies. We show that it is able to identify heterogeneous causal effects in MR by using it to study the causal relationship between high-density lipoprotein (HDL) cholesterol and coronary heart disease. In the third chapter, we introduce a variational inference algorithm for fitting robust Gaussian process regression (GPR) models with contaminated normal noise. Most existing robust GPR methods cannot be applied to large datasets as they perform exact inference which scales poorly with number of observations. We approximate the GPR model with contaminated normal noise with a sparse variational approximation and develop a generalized alternating minimization (GAM) algorithm for jointly estimating the variational and model parameters. Through simulation studies, we show that our pro

关键词： Applied statistics inference algorithms Probabilistic models Weather forecasting Space weather

来源：评论

学校读者我要写书评

暂无评论

FlashSchema: Achieving High Quality XML Schemas with Powerful inference algorithms and Large-scale Schema Data

FlashSchema: Achieving High Quality XML Schemas with Powerfu...

引用

International Conference on Data Engineering

作者： Yeting Li Jialun Cao Haiming Chen Tingjian Ge Zhiwu Xu Qiancheng Peng State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China The Hong Kong University of Science and Technology Hong Kong China University of Massachusetts Lowell United States College of Computer Science and Software Engineering Shenzhen University Shenzhen China

ISBN: (数字)9781728129037

ISBN: (纸本)9781728129044

Getting high quality XML schemas to avoid or reduce application risks is an important problem in practice, for which some important aspects have yet to be addressed satisfactorily in existing work. In this paper, we propose a tool FlashSchema for high quality XML schema design, which supports both one-pass and interactive schema design and schema recommendation. To the best of our knowledge, no other existing tools support interactive schema design and schema recommendation. One salient feature of our work is the design of algorithms to infer k-occurrence interleaving regular expressions, which are not only more powerful in model capacity, but also more efficient. Additionally, such algorithms form the basis of our interactive schema design. The other feature is that, starting from large-scale schema data that we have harvested from the Web, we devise a new solution for type inference, as well as propose schema recommendation for schema design. Finally, we conduct a series of experiments on two XML datasets, comparing with 9 state-of-the-art algorithms and open-source tools in terms of running time, preciseness, and conciseness. Experimental results show that our work achieves the highest level of preciseness and conciseness within only a few seconds. Experimental results and examples also demonstrate the effectiveness of our type inference and schema recommendation methods.

关键词： XML inference algorithms Tools Semantics Libraries Data models

来源：评论

学校读者我要写书评

暂无评论

A Model of Multi-Cascade Fuzzy Logic Controller Implemented Using Different Variations of inference algorithms

A Model of Multi-Cascade Fuzzy Logic Controller Implemented ...

引用

International Multi-Conference on Industrial Engineering and Modern Technologies (FarEastCon)

作者： S.P. Cherniy A.V. Buzikayeva A.S. Gudim Federal State-financed Educational Institution of Higher Professional Learning “Komsomolsk-on-Amur State Technical University” Komsomolsk-on-Amur Russia

In the present work, the directions for expanding the intellectual capabilities of fuzzy control systems of complex technological objects are shown. Often, the operation of the control object is carried out either in the conditions of incomplete information, or significant deficiencies in the mathematical description, both in terms of the system of differential equations and in the restrictions imposed on the system as a whole. Overcoming these shortcomings with the use of typical fuzzy approaches is fraught with difficulties associated both with the algorithmic complexity of the knowledge base, the information redundancy of the linguistic variables used, and a significant number of fuzzy terms in a given control range. In the implementation of generally accepted control procedures based on soft calculations, difficulties arise with the implementation of the adaptive properties of such systems. The combination of various algorithms of fuzzy inference allows solving intellectual control problems for multicriteria and multifactor problems. At the same time, a different combination of such fuzzy conclusions enhances the various properties of such systems, for example, robustness, multitasking, aggregate control in the areas of large and small signals, etc. The paper presents models of two systems with a combination of algorithms for fuzzy logic inference Sugeno-Sugeno and Sugeno-Mamdani, analyzes the main characteristics of control systems using such multi-stage systems, identifies advantages and disadvantages for various combinations of algorithms, presents simulation results. In addition, some difficulties and features of setting up an external cascade, based on Sugeno's fuzzy logic inference algorithm, were identified, which can later be eliminated through the use of alternative output algorithms, for example, Mamdani. Features of the application of an algorithm in the internal cascade are most often dictated by the specifics of the operation of the control object,

关键词： inference algorithms Fuzzy logic Process control Switches Industrial engineering Knowledge based systems

来源：评论

学校读者我要写书评

暂无评论

inference algorithms for the Multiplicative Mixture Mallows Model

Inference Algorithms for the Multiplicative Mixture Mallows ...

引用

International Conference on Signal Processing and Communications, SPCOM

作者： Ranjitha Prasad Vincent Y. F. Tan TCS Innovation Labs Delhi India National University of Singapore Singapore

ISBN: (纸本)9781538638224

A popular approach to obtain a consensus ranking from ranking data is based on the probabilistic, distance-based Mallows model comprising of a modal permutation and dispersion parameters. Often, the population consists of several subpopulations. As a result, finite mixture models are used to distinguish latent sub-groups of individuals in a heterogeneous population. Given a finite number of subpopulations each based on the Mallows model, a popular inference approach is the computationally intensive expectation maximization algorithm for additive models. We address the drawbacks of this model using a novel multiplicative mixture Mallows model (M4). Given complete ranking observations from a heterogeneous population, we derive inference algorithms for the joint estimation of the parameters and the consensus rankings of the component distributions. We numerically validate the permutation estimation performance of the proposed algorithms on synthetic datasets. We also demonstrate the goodness-of-fit using the Bayesian information criterion and the integrated complete likelihood on the real-world APA and Sushi datasets.

关键词： Computational modeling inference algorithms Sociology Statistics Additives Data models Dispersion

来源：评论

学校读者我要写书评

暂无评论

Review of Trends in Topic Modeling Techniques, Tools, inference algorithms and Applications

Review of Trends in Topic Modeling Techniques, Tools, Infere...

引用

International Conference on Soft Computing and Machine Intelligence (ISCMI)

作者： Christine K. Mulunda Peter W. Wagacha Lawrence Muchemi School of Computing and Informatics University of Nairobi Nairobi Kenya

ISBN: (纸本)9781728113029;9781728113012

Topic modeling algorithms discover latent abstract topics in large volumes of data. The choice of inference algorithm determines the kind of solution the topic models will provide. inference algorithms aim at fast convergence of topics and stability of solutions with respect to initial parameters. The problem area identified for a topic model forms the basis of selecting the inference algorithm. This study reviewed 85 articles from various journals, classified them as reviews, techniques, tools, inference algorithms for topic models and application areas, among others. Scalability, computation speed and dynamism of topic models are necessary for robust solutions in analysis of datasets that are voluminous, varied and that scales up very fast with time.

关键词： Computational modeling inference algorithms Data models Tools Analytical models Real-time systems Approximation algorithms

来源：评论

学校读者我要写书评

暂无评论

Offloading algorithms for Maximizing inference Accuracy on Edge Device in an Edge Intelligence System

引用

IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS 2023年第7期34卷 2025-2039页

作者： Fresa, Andrea Champati, Jaya Prakash IMDEA Networks Inst Edge Networks Grp Madrid 28918 Spain

With the emergence of edge computing, the problem of offloading jobs between an Edge Device (ED) and an Edge Server (ES) received significant attention in the past. Motivated by the fact that an increasing number of applications are using Machine Learning (ML) inference from the data samples collected at the EDs, we study the problem of offloading inference jobs by considering the following novel aspects: 1) in contrast to a typical computational job, the processing time of an inference job depends on the size of the ML model, and 2) recently proposed Deep Neural Networks (DNNs) for resource-constrained devices provide the choice of scaling down the model size by trading off the inference accuracy. Considering that multiple ML models are available at the ED, and a powerful ML model is available at the ES, we formulate an Integer Linear Programming (ILP) problem with the objective of maximizing the total inference accuracy of n data samples at the ED subject to a time constraint T on the makespan. Noting that the problem is NP-hard, we propose an approximation algorithm Accuracy Maximization using LP-Relaxation and Rounding (AMR(2)) and prove that it results in a makespan at most 2T and achieves a total accuracy that is lower by a small constant from the optimal total accuracy implying that AMR(2) is asymptotically optimal. Further, if the data samples are identical we propose Accuracy Maximization using Dynamic Programming (AMDP), an optimal pseudo-polynomial time algorithm. Furthermore, we extend AMR(2) for the case of multiple ESs, where each ES is equipped with a powerful ML model. As proof of concept, we implemented AMR(2) on a Raspberry Pi, equipped with MobileNets, that is connected to a server equipped with ResNet, and studied the total accuracy and makespan performance of AMR(2) for image classification.

关键词： Data models Computational modeling inference algorithms Costs Servers Approximation algorithms Scheduling deep learning edge computing linear programming machine learning

来源：评论

学校读者我要写书评

暂无评论

An Attack-Defense Game-Based Reinforcement Learning Privacy-Preserving Method Against inference Attack in Double Auction Market

引用

IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING 2025年 22卷 9075-9089页

作者： Li, Donghe Hu, Chunlin Yang, Qingyu Ma, Yuhao Zhang, Feiye An, Dou Xi An Jiao Tong Univ Sch Automat Sci & Engn Xian 710049 Peoples R China Xi An Jiao Tong Univ MOE Key Lab Intelligent Networks & Network Secur Xian 710049 Peoples R China Xi An Jiao Tong Univ SKLMSE Lab Xian 710049 Peoples R China Xi An Jiao Tong Univ Sch Automat Sci & Engn Xian 710049 Peoples R China

Auction mechanism, as a fair and efficient resource allocation method, has been widely used in varieties trading scenarios, such as advertising, crowdsensoring and spectrum. However, in addition to obtaining higher profits and satisfaction, the privacy concerns have attracted researchers' attention. In this paper, we mainly study the privacy preserving issue in the double auction market against the indirect inference attack. Most of the existing works apply differential privacy theory to defend against the inference attack, but there exists two problems. First', indistinguishability' of differential privacy (DP) cannot prevent the disclosure of continuous valuations in the auction market. Second, the privacy-utility trade-off (PUT) in differential privacy deployment has not been resolved. To this end, we proposed an attack-defense game-based reinforcement learning privacy-preserving method to provide practically privacy protection in double auction. First, the auctioneer acts as defender, adds noise to the bidders' valuations, and then acts as adversary to launch inference attack. After that the auctioneer uses the attack results and auction results as a reference to guide the next deployment. The above process can be regarded as a Markov Decision Process (MDP). The state is the valuations of each bidders under the current steps. The action is the noise added to each bidders. The reward is composed of privacy, utility and training speed, in which attack success rate and social welfare are taken as measures of privacy and utility, a delay penalty term is used to reduce the training time. Utilizing the deep deterministic policy gradient (DDPG) algorithm, we establish an actor-critic network to solve the problem of MDP. Finally, we conducted extensive evaluations to verify the performance of our proposed method. The results show that compared with other existing DP-based double auction privacy preserving mechanisms, our method can achieve better results in both privacy

关键词： Privacy Protection Differential privacy Noise Games Reinforcement learning Cost accounting Automation Training inference algorithms Double auction inference attack differential privacy privacy-utility trade-off reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Joint Optimization of Device Placement and Model Partitioning for Cooperative DNN inference in Heterogeneous Edge Computing

引用

IEEE TRANSACTIONS ON MOBILE COMPUTING 2025年第1期24卷 210-226页

作者： Dai, Penglin Han, Biao Li, Ke Xu, Xincao Xing, Huanlai Liu, Kai Southwest Jiaotong Univ Sch Comp & Artificial Intelligence Chengdu 611756 Peoples R China Univ Elect Sci & Technol China Shenzhen Inst Adv Study Chengdu 610051 Peoples R China Chongqing Univ Coll Comp Sci Chongqing 400040 Peoples R China

EdgeAI represents a compelling approach for deploying DNN models at network edge through model partitioning. However, most existing partitioning strategies have primarily concentrated on homogeneous environments, neglecting the effect of device placement and their inapplicability to heterogeneous settings. Moreover, these strategies often rely on either data parallelism or model parallelism, each presenting its own limitations, including data synchronization and communication overhead. This paper aims at enhancing inference performance through a pipeline system of devices through leveraging both parallel and sequential relationships among them. Accordingly, the problem of Multi-Device Cooperative DNN inference is formulated by optimizing both device placement and model partitioning, taking into account the unique characteristics of heterogeneous edge resources and DNN models, with the goal of maximizing throughput. To this end, we propose an evolutionary device placement technique to determine the pipeline stage of devices by enhancing a variant of particle swarm optimization. Subsequently, an adaptive model partitioning strategy is developed by combining intra-layer and inter-layer model partitioning based on dynamic programming and the input-output mapping of DNN layers, respectively, to accommodate edge resource limitations. Finally, we construct a simulation model and a prototype, and the extensive results demonstrate that our proposed algorithm outperforms current state-of-the-art algorithms.

关键词： Computational modeling Artificial neural networks Parallel processing Data models Adaptation models Partitioning algorithms inference algorithms Multi-device cooperative inference heterogeneous edge computing DNN Model Partitioning device placement

来源：评论

学校读者我要写书评

暂无评论

Distributed Collaborative inference System in Next-Generation Networks and Communication

引用

IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING 2025年第2期11卷 923-932页

作者： Zhang, Chuan Zheng, Xixi Tao, Xiaolong Hu, Chenfei Zhang, Weiting Zhu, Liehuang Beijing Inst Technol Sch Cyberspace Sci & Technol Beijing 100081 Peoples R China Minist Educ China Key Lab Cyberspace Secur Zhengzhou 450001 Peoples R China Beijing Jiaotong Univ Sch Elect & Informat Engn Beijing 100044 Peoples R China

With the rapid advancement of artificial intelligence, generative artificial intelligence (GAI) has taken a leading role in transforming data processing methods. However, the high computational demands of GAI present challenges for devices with limited resources. As we move towards the sixth generation of mobile networks (6G), the higher data rates of 6G create a need for more efficient data processing in GAI. Traditional GAI, however, shows its limitations in meeting these demands. To address these challenges, we introduce a multi-level collaborative inference system designed for next-generation networks and communication. Our proposed system features a deployment strategy that assigns models of varying sizes to devices at different network layers. Then, we design a task offloading strategy to optimise both efficiency and latency. Furthermore, a modified early exit mechanism is implemented to enhance the inference process for single models. Experimental results demonstrate that our system effectively reduces inference latency while maintaining high-quality output. Specifically, compared to existing work, our system can reduce inference time by up to 17% without sacrificing the inference accuracy.

关键词： Computational modeling Accuracy Transformers inference algorithms Collaboration Next generation networking Adaptation models 6G mobile communication Servers Probabilistic logic Generative artificial intelligence early exit collaborative inference next-generation networks and communication

来源：评论

学校读者我要写书评

暂无评论

A 4.69-TOPS/W Training, 2.34-μJ/Image inference On-Chip Training Accelerator With inference-Compatible Backpropagation and Design Space Exploration in 28-nm CMOS

引用

IEEE JOURNAL OF SOLID-STATE CIRCUITS 2025年第1期60卷 298-307页

作者： Qian, Junyi Ge, Haitao Lu, Yicheng Shan, Weiwei Southeast Univ Sch Integrated Circuits Nanjing 210096 Peoples R China

On-chip training (OCT) accelerators improve personalized recognition accuracy while ensuring user privacy. However, previous OCT accelerators often required significant additional hardware costs to support retraining, even though inference is the primary use case. We propose an inference-pattern-compatible backpropagation (BP) circuit, which enables the training process to reuse inference hardware. To achieve high energy efficiency, we utilize three hardware-friendly optimization methods that significantly reduce redundant computation and external memory access (EMA). Additionally, we propose a design space exploration (DSE) to search for the optimal hardware configurations, which improves system performance while reducing the design time cost. Fabricated in a 28-nm CMOS process, this single-core OCT chip is able to train all the layers of a neural network (NN), achieving a peak training efficiency of 4.69 Tera operations per second per watt (TOPS/W). It also achieves the lowest inference energy of 2.34 mu J/inf/image under a core voltage of 0.48 V and 40 MHz.

关键词： Convolution Training Kernel Hardware inference algorithms System-on-chip Energy efficiency Backpropagation (BP) circuit design space exploration (DSE) energy efficiency neural network (NN) on-chip training (OCT) transfer learning

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：