检索结果-内蒙古大学图书馆

Byzantine Resilient Non-Convex SCSG With Distributed Batch Gradient Computations

IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS 2021年 7卷 754-766页

作者： Bulusu, Saikiran Khanduri, Prashant Kafle, Swatantra Sharma, Pranay Varshney, Pramod K. Syracuse Univ Dept EECS Syracuse NY 13244 USA Univ Minnesota Dept ECE Minneapolis MN 55455 USA

Distributed learning is an important paradigm in the current machine learning algorithms with large datasets. In this paper, distributed stochastic optimization problem of minimizing a nonconvex function in an adversarial setting is considered. A robust variant of the stochastic variance-reduced algorithm is proposed. In the distributed setup, we assume that a fraction of worker nodes (WNs) can be Byzantines. We assume that the batch gradients are computed at the WNs and the stochastic gradients are computed at the central node (CN). We provide the convergence rate of the proposed algorithm which employs the design of a novel filtering rule that is independent of the problem dimension. Furthermore, we capture the effect of Byzantines present in the network on the convergence performance of the algorithm. We evaluate the performance of the proposed algorithm and present the simulation results using real world datasets, in addition to providing the theoretical guarantees.

关键词： Convergence Linear programming Optimization machine learning algorithms Distance learning Computer aided instruction Stochastic processes Nonconvex optimization distributed optimization variance reduction Stochastic gradient descent

来源：评论

学校读者我要写书评

暂无评论

Indoor Positioning System Using Synthetic Training and Data Fusion

引用

IEEE ACCESS 2021年 9卷 115687-115699页

作者： Assayag, Yuri Oliveira, Horacio Souto, Eduardo Barreto, Raimundo Pazzi, Richard Univ Fed Amazonas Inst Comp BR-69067005 Manaus Amazonas Brazil Ontario Tech Univ UOIT Fac Business & Informat Technol Oshawa ON L1H 7K4 Canada

Indoor Positioning Systems (IPSs) are used to estimate the position of mobile devices in indoor environments. Fingerprinting is the most used technique because of its higher accuracy. However, this technique requires a labor-intensive training phase that measures the Received Signal Strength Indicator (RSSI) at all Reference Points (RPs) locations. On the other hand, model-based IPSs use signal propagation models to estimate distances from RSSI. Thus, they do not require expensive training but result in higher positioning errors. In this work, we propose SynTra-IPS (Synthetic Training Indoor Positioning System), a hybrid approach between a fingerprint and a model-based IPS that uses synthetic, simulated datasets combined with data fusion techniques to eliminate the fingerprint collection cost. In our solution, we use the map of the scenario, with known anchor nodes' positions and the log-distance signal propagation model, to generate several synthetic, model-based, fingerprint training datasets. In the online phase of our solution, the positions estimated by the several synthetic datasets using K-Nearest Neighbors (KNN) are combined using data fusion techniques into a single, more accurate position. We evaluated the performance of our SynTra solution in a real-world, large-scale environment using mobile devices with Bluetooth Low Energy (BLE) technology, and we compared our solution to classic approaches from the literature. Our results show that SynTra can locate mobile devices with an average error of only 2.36 m while requiring no real-world environment training.

关键词： Training Mobile handsets IP networks Data integration Location awareness machine learning algorithms Indoor environment Indoor localization fingerprint path-loss propagation model synthetic dataset

来源：评论

学校读者我要写书评

暂无评论

Few-Shots Parallel Algorithm Portfolio Construction via Co-Evolution

引用

IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION 2021年第3期25卷 595-607页

作者： Tang, Ke Liu, Shengcai Yang, Peng Yao, Xin Southern Univ Sci & Technol Dept Comp Sci & Engn Guangdong Key Lab Brain Inspired Intelligent Comp Shenzhen 518055 Peoples R China Southern Univ Sci & Technol Res Inst Trustworthy Autonomous Syst Shenzhen 518055 Peoples R China

Generalization, i.e., the ability of solving problem instances that are not available during the system design and development phase, is a critical goal for intelligent systems. A typical way to achieve good generalization is to learn a model from vast data. In the context of heuristic search, such a paradigm could be implemented as configuring the parameters of a parallel algorithm portfolio (PAP) based on a set of "training" problem instances, which is often referred to as PAP construction. However, compared to the traditional machine learning, PAP construction often suffers from the lack of training instances, and the obtained PAPs may fail to generalize well. This article proposes a novel competitive co-evolution scheme, named co-evolution of parameterized search (CEPS), as a remedy to this challenge. By co-evolving a configuration population and an instance population, CEPS is capable of obtaining generalizable PAPs with few training instances. The advantage of CEPS in improving generalization is analytically shown in this article. Two concrete algorithms, namely, CEPS-TSP and CEPS-VRPSPDTW, are presented for the traveling salesman problem (TSP) and the vehicle routing problem with simultaneous pickup-delivery and time windows (VRPSPDTW), respectively. The experimental results show that CEPS has led to better generalization, and even managed to find new best-known solutions for some instances.

关键词： Training Portfolios Tuning Statistics Sociology Parallel algorithms machine learning algorithms Algorithm configuration automatic parameter tuning co-evolution parallel algorithm portfolios (PAPs) vehicle routing problems

来源：评论

学校读者我要写书评

暂无评论

Recognition of Trip-Based Aggressive Driving: A System Integrated With Gaussian Mixture Model Structured of Factor-Analysis, and Hierarchical Clustering

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2022年第11期23卷 20442-20451页

作者： Wang, Junhua Xu, Wenxiang Fu, Ting Jiang, Rui Tongji Univ Key Lab Rd & Traff Engn Minist Educ Shanghai 201804 Peoples R China China Shandong Int Econ & Tech Cooperat Grp Ltd Investment & Dev Dept Jinan 250098 Shandong Peoples R China

Recognition of aggressive driving behavior helps future research and practices in Intelligent Transport Systems. This paper tries to briefly explain the concepts related to aggressive driving behavior and introduces a system integrated with a novel machine learning algorithm for the recognition of trip-based aggressive driving behavior. The algorithm is a Gaussian Mixture Model (GMM) structured with Factor Analysis (FA), and Hierarchical Clustering (HC): common factors were extracted using FA, which is further applied to HC and GMM in the recognition of trip-based aggressive driving. The system is applied in a case study using data from the Shanghai Naturalistic Driving Study, for simulating data collection using the Advanced Driving Assistance System (ADAS) system in a real-traffic situation. Three behavior types (cautious, regular, and aggressive driving) were successfully clustered. For validity, the real aggressive driving behavior records were extracted based on the video, and the proposed system was compared with existing recognition methods. Results indicate that the accuracy of aggressive driving recognition of the system is higher than others (accuracy = 87%). This paper provides a reference in defining and determining aggressive driving, and a robust system for aggressive driving behavior recognition along with the trained algorithm, which can be used in real-world applications for improving driving safety with the applications in ADAS systems, auto-insurance industry.

关键词： Behavioral sciences Vehicles Data models Clustering algorithms Computational modeling machine learning algorithms Dimensionality reduction Aggressive driving driving behavior mixture model hierarchical clustering Gaussian mixture model

来源：评论

学校读者我要写书评

暂无评论

Forecasting Electricity Price in Different Time Horizons: An Application to the Italian Electricity Market

引用

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS 2021年第6期57卷 5726-5736页

作者： Imani, Mahmood Hosseini Bompard, Ettore Colella, Pietro Huang, Tao Politecn Torino Dipartimento Energia I-10129 Turin Italy

Electricity price is a crucial element for market players to maximize their profits. In this context, the forecast of the hour-ahead, day-ahead, and week-ahead electricity prices plays a crucial role. The more accurate the prediction is, the lower the market risk is. In this article, several machine learning algorithms (support vector machine, Gaussian processes regression, regression trees, and multilayer perceptron) with different structures have been adopted to forecast Italian wholesale electricity prices. Considering different time horizons (hourly, daily, and weekly), their performances have been compared through several performance metrics, including mean absolute error, R-index, mean absolute percentage error, and the number of anomalies in which the forecast error passes a threshold. The investigation reveals that, in general, SVM and tree-based models outperform other models at different time horizons.

关键词： Support vector machines Prediction algorithms Forecasting Regression tree analysis Predictive models machine learning algorithms Artificial neural networks Different forecasting horizons electricity price prediction Italian electricity market machine learning (ML) prediction error distribution PUN

来源：评论

学校读者我要写书评

暂无评论

Interrelate Training and Clustering for Online Speaker Diarization

引用

IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 2024年 32卷 1352-1364页

作者： Chen, Yifan Cheng, Gaofeng Yang, Runyan Zhang, Pengyuan Yan, Yonghong Chinese Acad Sci Inst Acoust Key Lab Speech Acoust & Content Understanding Beijing 100190 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China

In clustering-based speaker diarization systems, the embedding clusters for distinctive speakers exhibit wide variability in size and density, posing difficulty for clustering accuracy. In spite of this, with the assistance of the overall distance relationships among speaker embeddings, most of the embeddings can be grouped to the correct cluster by sophisticated offline clustering algorithms. However, in online scenarios, such a complete distance relationships of the embeddings can not be obtained due to the incremental arrival of embeddings. Consequently, determining the number of clusters and then correctly grouping the embeddings become challenging in an online fashion. Furthermore, errors would accumulate quickly over time if the online clustering algorithm assigns the embeddings into clusters erroneously in the beginning. To address these problems, we designed a novel framework for online clustering. To reduce the high variability of speaker embeddings, we proposed the clustering guided embedding extractor training (CGEET) algorithm to encourage similarity between the size of the embedding space for different speakers in attempt to simplify the distance relationships of embeddings. The CGEET algorithm can grasp the distance information of the entire speaker embedding space and provide it to the online clustering algorithm. With this preliminary information, the distance thresholds guided online clustering (DTGOC) algorithm then processes incoming embeddings using a divide-and-conquer approach. It first handles the embeddings with explicit distance relationships and then searches for possible path combination they have with remaining embeddings in an online fashion. Moreover, in order to utilize the distance relationships of embeddings that are far apart in time, an online re-clustering strategy is incorporated in our DTGOC algorithm, which can alleviate error accumulation during online clustering. By implementing the above innovations, our proposed online clus

关键词： Clustering algorithms Training Task analysis Oral communication Speech processing Acoustics machine learning algorithms Metric embedding learning online clustering speaker diarization

来源：评论

学校读者我要写书评

暂无评论

SDTR: Soft Decision Tree Regressor for Tabular Data

引用

IEEE ACCESS 2021年 9卷 55999-56011页

作者： Luo, Haoran Cheng, Fan Yu, Heng Yi, Yuqi Shanghai Jiao Tong Univ AI Inst Dept Comp Sci & Engn MoE Key Lab Artificial Intelligence Shanghai 200240 Peoples R China

Deep neural networks have been proved a success in multiple fields. However, researchers still favor traditional approaches to obtain more interpretable models, such as Bayesian methods and decision trees when processing heterogeneous tabular data. Such models are hard to differentiate, thus inconvenient to be integrated into end-to-end settings. On the other hand, traditional neural networks are differentiable but perform poorly on tabular data. We propose a hierarchical differentiable neural regression model, Soft Decision Tree Regressor (SDTR). SDTR imitates a binary decision tree by a differentiable neural network and is plausible for ensemble schemes like bagging and boosting. The SDTR method was evaluated on multiple tabular-based regression tasks (YearPredictionMSD, MSLR-Web10K, Yahoo LETOR, SARCOS and Wine quality). Its performance is comparable with non-differentiable models (gradient boosting decision trees) and better than uninterpretable models (regular FCNN). On top of that, it can produce fair results with a restricted number of parameters, only using a small forest or even a single tree. We also propose an "average entropy" metric to evaluate the level of interpretability of a trained, soft decision tree neural network. This metric also helps to select proper structure and hyperparameters for such networks.

关键词： Regression tree analysis Neural networks Vegetation Task analysis Data models Boosting Measurement Decision trees neural networks machine learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Factual and Counterfactual Explanations for Black Box Decision Making

引用

IEEE INTELLIGENT SYSTEMS 2019年第6期34卷 14-22页

作者： Guidotti, Riccardo Monreale, Anna Giannotti, Fosca Pedreschi, Dino Ruggieri, Salvatore Turini, Franco ISTI CNR Knowledge Discovery & Data Min Lab Pisa Italy Univ Pisa Comp Sci Dept Pisa Italy ISTI CNR Res Pisa Italy Univ Pisa Comp Sci Pisa Italy Univ Pisa Pisa Italy

The rise of sophisticated machine learning models has brought accurate but obscure decision systems, which hide their logic, thus undermining transparency, trust, and the adoption of artificial intelligence (AI) in socially sensitive and safety-critical contexts. We introduce a local rule-based explanation method, providing faithful explanations of the decision made by a black box classifier on a specific instance. The proposed method first learns an interpretable, local classifier on a synthetic neighborhood of the instance under investigation, generated by a genetic algorithm. Then, it derives from the interpretable classifier an explanation consisting of a decision rule, explaining the factual reasons of the decision, and a set of counterfactuals, suggesting the changes in the instance features that would lead to a different outcome. Experimental results show that the proposed method outperforms existing approaches in terms of the quality of the explanations and of the accuracy in mimicking the black box.

关键词： Genetic algorithms Intelligent systems Decision making Decision trees machine learning algorithms Prediction algorithms Data models Explainable AI Interpretable machine learning Open the Black Box Explanation Rules Counterfactuals

来源：评论

学校读者我要写书评

暂无评论

An Improved Nonparallel Support Vector machine

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND learning SYSTEMS 2021年第11期32卷 5129-5143页

作者： Liu, Liming Chu, Maoxiang Gong, Rongfen Zhang, Li Univ Sci & Technol Liaoning Sch Elect & Informat Engn Anshan 114051 Peoples R China

In this article, an improved nonparallel support vector machine (INPSVM) is proposed for pattern classification. INPSVM inherits almost all advantages of nonparallel support vector machine (NPSVM), i.e., the kernel trick can be directly applied for the nonlinear case and the matrix inversion is avoided. These are completely different from the twin support vector machine (TSVM). Moreover, the INPSVM classifier has some incomparable advantages over TSVM and NPSVM. First, it can effectively eliminate the negative effect of noise, especially feature noise around the decision boundary. Second, the novel classifier has higher classification accuracy for both linear and nonlinear data sets compared with the other algorithms. Finally, a large number of experiments show that INPSVM is superior to other algorithms in efficiency, accuracy, and robustness.

关键词： Support vector machines Training machine learning algorithms Fasteners learning systems Kernel machine learning Generalization performance noise insensitivity nonparallel support vector machine (NPSVM) pattern classification twin support vector machine (TSVM)

来源：评论

学校读者我要写书评

暂无评论

Chemometrics for QSAR with low sequence homology:: Mycobacterial promoter sequences recognition with 2D-RNA entropies

引用

CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS 2007年第1期85卷 20-26页

作者： Gonzalez-Diaz, Humberto Perez-Bello, Alcides Cruz-Monteagudo, Maykel Gonzalez-Diaz, Yenny Santana, Lourdes Uriarte, Eugenio Univ Santiago de Compostela Fac Pharm Dept Organ Chem Santiago De Compostela 15782 Spain Provincial Ctr Human Genet Las Tunas 77400 Cuba ICBP Victoria Giron Natl Ctr Human Genet Havana 11600 Cuba

Predicting mycobacterial sequences promoter of protein synthesis is important in the study of protein metabolism regulation. This goal is however considered a challenging computational biology task due to low inter-sequences homology. Consequently, a previous work based only on DNA sequence had to use a large input parameter set and multilayered feed-forward ANN architecture trained using the error-back-propagation algorithm to raise an overall accuracy up to 97% [Kalate, et al. 2003. Comput. Biol. Chem. 27, 555-564]. Subsequently, one could expect that a notably simpler model may be derived using parameters based on non-linear structural information. In the present work, a method based on molecular folding negentropies (Theta(k)) is introduced to predict by the first time mycobacterial promoter sequences (mps) from the corresponding RNA secondary structure. The best QSAR equation found was the classification function mps = 4.921 x (0)Theta(M) - 1.205, which recognised 126/135 mps (93.3%) and 100% of 245 control sequences (cs). The model have shown a very high Mathew regression coefficient C = 0.949. Both average overall accuracy and predictability were 97.6%. Additionally, several machine learning algorithms were applied in order to reaffirm the validity of the LDA model from the chemometrics point of view. This linear model with only one parameter ((0)Theta(M)) may be considered the simpler reported up-to-date by large, without lack of accuracy (97%) with respect to Kalate et al.'s model. (c) 2006 Elsevier B.V. All rights reserved.

关键词： mycobacterial promoter sequences RNA secondary structure Markov models machine learning algorithms QSAR information theory entropy

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：