查看更多>>摘要：In this paper, we propose a novel method for 3D human joint estimation using part seg-mentation, and introduce an application for size measurement based on the obtained joints. A human segmentation dataset is first prepared as training set for the advanced neu-ral network architecture. Different human parts yielded from the neural network are uti-lized to extract human joints. In the proposed method, the joints are categorized into the active joints and inert joints. In the extraction process of the active joints, the mathe-matical analysis method is adopted to calculate the joint positions. The geometric features of different human segments are further used to extract the inert joints. Moreover, we test on the dataset to compare its performance with our previous method based on geometrical features. The results show the average error of the joints is less than 4.2 cm, which is sig-nificantly improved from 5.8 cm demonstrated in our previous research. We also investi-gate the human size measurement. The distance between the joints is used to calculate the length, and the ellipse fitting method based on multi-frame point cloud is adapted to calculate the human girths. Compared with the manual measurement data, the size error is less than 4.1 cm.(c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：This paper is concerned with the consensus tracking problem of stochastic multi-agent systems with both output, partial state constraints, and input saturation via event-triggered strategy. To handle with the saturated control inputs, the saturation function is transformed into a linear form of the control input. By using radial basis function neural network to approximate the unknown nonlinear function, the unmeasurable states are acquired by an adaptive observer. To ensure that the constraints of system outputs and partial states are never violated, an appropriate time-varying barrier Lyapunov function is constructed. The control scheme is event-triggered in order to save communication resources. The proposed distributed controller can guarantee the boundedness of all system signals, the consensus tracking with a small bounded error, and the avoidance of the Zeno behavior by using backstepping techniques. The validity of the theoretical results is verified by computer simulation.
Bustio-Martinez, LazaroAlvarez-Carmona, Miguel A.Herrera-Semenets, VitaliFeregrino-Uribe, Claudia...
18页
查看更多>>摘要：Phishing is a cyber-attack that exploits victims' technical ignorance or naivety and commonly involves a Uniform Resources Locator (URL). Hence, it is advantageous to detect a phishing attack by analyzing URLs before accessing them. With the raising of the Internet of Things (IoT), phishing attacks are moving to this field due to the number of IoT devices and the amount of personal information they handle. Although several approaches were proposed for phishing attacks detection, the URL-based Machine Learning approaches obtain better performance results, but all of them are dependent on the feature set used. Contradictorily, only a few works on selecting the best-suited feature set for improving the phishing detection process have been published. The present research explores how to obtain a feature set that substantially enhances the phishing detection rate in IoT environments. Hence, a feature selection algorithm was adopted and extended for getting the most representative feature set. When Random Forest is used with the proposed data representation, the phishing URL attacks discovery rate is 99.57%. (c) 2022 Published by Elsevier Inc.
查看更多>>摘要：Multi-attribute decision-making (MADM) aims to rank alternatives based on their attri-butes and evaluation information, which provides decision support for decision-makers. Most existing MADM methods can only obtain ranking results, decision-makers usually need to subjectively choose priority alternatives to make a decision based on the preset evaluation level and ranking results, which does not meet the requirements of complex decision situations and uncertain information processing. A method is urgently needed, which can objectively produce classification results and automatically provide priority objects for decisions. Three-way decisions (3WDs) offer an effective research technique to solve decision-making problems under uncertainty and risk by objectively dividing a universal set into three pairwise disjoint parts: acceptance, non-commitment and rejec-tion, and implementing the corresponding strategy for each part. Considering that it is dif-ficult to accurately determine the exact value of attributes in some cases, interval numbers can be a useful concept to flexibly describe uncertain information and satisfy the decision-maker's cognition. Therefore, under an interval-valued MADM environment, this paper proposes an evaluation-based interval-valued multi-attribute three-way decision (IVMA3WD) model from an optimization viewpoint. The model achieves an effective fusion of 3WDs and MADM problems, and further expands evaluation-based 3WD models on the totally ordered set. Specifically speaking, in order to depict uncertain information in decision-making, interval numbers are used to describe the evaluation information included in the multi-attribute evaluation matrix. Then, considering a general case that there are no decision attributes in the multi-attribute evaluation matrix, we determine the evaluation function by use of the idea of the interval-valued TOPSIS method. In addi-tion, unlike the fixed-valued loss functions in the majority of 3WD models, the relative loss function in this article is not provided straightly but computed by attribute evaluation val-ues expressed by interval numbers. Furthermore, based on an optimization viewpoint, we construct two optimization models and attempt to find the optimal solution to obtain the threshold pair. A housing market investment decision problem is utilized as an illustrative example to find out the applicability of our built IVMA3WD model. In the end, we adopt a sensitivity analysis and a comparative analysis to demonstrate our proposed model's char-acteristics and advantages.(c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：This article presents an adaptive neural network event-triggered asymptotic tracking control strategy for multi-input and multi-output (MIMO) nonlinear systems with state constraints and unknown dynamics. The prescribed state constraints are ensured by employing barrier Lyapunov function and the neural networks are utilized to address unknown dynamics. By applying a bound estimation method and some smooth functions, associating with backstepping technique, the asymptotic tracking controller is recursively constructed. Meanwhile, the event-triggered mechanism is introduced into the design process to mitigate data transmission. Finally, the validity of the presented asymptotic controller is elucidated via a practical simulation example. (c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：In this paper, a novel approach to the self-organization of hierarchical prototype-based classifiers from data is proposed. The approach recursively partitions the data at multiple levels of granularity into shape-free clusters of different sizes, resembling Voronoi tessellation, and naturally aggregates the resulting cluster medoids into a multi-layered prototype-based structure according to their descriptive abilities. Different from conventional classification models, it is nonparametric and entirely data-driven, and the learned model can offer a high-level of transparency and interpretability thanks to the underlying prototype-based nature. The system identification process underpinning the approach is driven by the aim of separating data samples of different classes into nonoverlapping multi-granular clusters. Its associated decision-making process follows the "nearest prototype" principle and hence, the rationales of the subsequent decisions made can be explicitly explained. Experimental studies based on popular benchmark classification problems, as well as on a practical application to remote sensing image classification, demonstrate the efficacy of the proposed approach.(c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Many network-based tasks need powerful feature expression to capture the diversity of networks. It can be provided by network embedding learning of nodes. The related researches led to a significant progress so far. Nevertheless, suffering from paying unique attention on the structure or neighborhoods of nodes, most of methods cannot describe the vector representations of nodes well enough. In addition, random walk plays a key role during the learning procedure in many advanced methods. The successor node is selected according t the proximity of its direct prior node in each walk step. It is easy to lead a similarity drifting in a long walk sequence and influence the sequence quality. To address these issues, a fuzzy hierarchical network embedding approach (FHNE) is put forward to learn the vector expression fusing structure and neighbor information. Firstly, a multi-granular graph is constructed by a fuzzy k-core decomposition to encode the structural and neighborhood information of nodes. Then, inspired by the epidemic method, a biased random walk is designed to solve the similarity drifting. Many numerical experiments demonstrate that our method exhibits superior performance on various tasks in some real datasets. It verifies that FHNE can learn the efficient network representations. (c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Design alternative assessment is vital in product development since it directly influences the directions of subsequent design and manufacturing activities. The alternative assessment information chiefly depends on experts' subjective perceptions and preferences, which include several types of uncertainty, such as intrapersonal perception ambiguousness, personal judgment reliability, and interpersonal preference inconsistency. However, previous studies usually just consider one of the various uncertainties, which may affect their effectiveness. To fill this gap, we develop an integrated design alternative assessment model integrating Z-cloud rough numbers (ZCRNs), best-worst method (BWM), and multi attributive border approximation area comparison (MABAC). First, to fully handle various uncertainties, a new concept of ZCRN is developed by combining the benefits of cloud model in addressing intrapersonal uncertainty, the merits of Z-numbers in reflecting judgmental reliability, and the strengths of rough numbers in handling interpersonal uncertainty. Some arithmetic operating rules, Minkowski-type distance, comparison measure, correlation measure, and aggregation operators for ZCRNs are also introduced. Furthermore, a ZCRN-BWM method and a ZCRN-MABAC method are developed to calculate criteria weights and rank design alternatives. Finally, a case study, sensitivity analysis on two parameters and a normalization method, and several comparisons are performed to elaborate and validate the developed model.(c) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：With the advance of machine learning technology and especially the explosive growth of big data, federated learning, which allows multiple participants to jointly train a high-quality global machine learning model, has gained extensive attention. However, in federated learning, it has been proved that inference attacks could reveal sensitive information from both local updates and global model parameters, which threatens user privacy greatly. Aiming at the challenge, in this paper, a privacy-preserving and lossless federated learning scheme, named CORK, is proposed for deep neural network. With CORK, multiple participants can train a global model securely and accurately with the assistance of an aggregation server. Specifically, we first design a drop-tolerant secure aggregation algorithm FTSA, which ensures the confidentiality of local updates. Then, a lossless model perturbation mechanism PTSP is proposed to protect sensitive data in global model parameters. Furthermore, the neuron pruning operation in PTSP can reduce the scale of models, which thus improves the computation and communication efficiency significantly. Detailed security analysis shows that CORK can resist inference attacks on both local updates and global model parameters. In addition, CORK is implemented with real MNIST and CIFAR-10 datasets, and the experimental results demonstrate that CORK is indeed effective and efficient. (c) 2022 Elsevier Inc. All rights reserved.
Liu, LichengCai, LuyangLiu, TingyunChen, C. L. Philip...
12页
查看更多>>摘要：Recently, the Broad Learning System (BLS) has attracted much attention due to its efficiency in clean data regression tasks. However, the conventional BLS performs poorly in the noisy environment, because the least square regression based loss function it used for networking training is noise sensitive. To address this problem, the Cauchy Regularized BLS (CRBLS) is proposed in this paper for noisy data prediction. Specifically, the Cauchy loss function instead of the least square metric is introduced into the system to regularize the residues. Compared to the least square loss, the Cauchy loss can penalize the large noise terms owe to its nonlinearity and bounded influence function. This admits the proposed CRBLS to handle data corrupted by Gaussian noise and outliers with different noise levels. In addition, a new incremental learning algorithm is developed for fast model updating without retraining the whole network when additional training samples are added. To evaluate the feasibility of CRBLS, extensive experiments are conducted, and results show that the proposed CRBLS achieves robustness for noisy data predication. (c) 2022 Elsevier Inc. All rights reserved.
NSTL
Elsevier