查看更多>>摘要：Blockchain-enabled distributed networks present a new infrastructure for building reliable and privacy-preserving healthcare utilities. Among them, Ethereum networks have gained specialized attention for their high usability and security. In this paper, we raise the awareness of Ethereum's vulnerability due to selfish mining, in which ill-disposed miners are feasible to receive more rewards than honest ones. To demonstrate this, we compose a new bribery selfish mining scheme, the BSM-Ether, targeted to Ethereum. The BSM-Ether attack can be easily conducted and has higher rewards than other existing malicious attacks. Besides, we present a formal and rigorous analysis of multiple kinds of rewards from the BSM-Ether attack. Simulation experiments show the high effectiveness of BSM-Ether the attacker can get a higher revenue with few efforts. To tackle this variability, we present some implementation level proposals and suggestions for building healthcare systems on top of Ethereum to minimize the negative effect of the security deficiency of underlying systems. (C) 2022 Published by Elsevier Inc.
查看更多>>摘要：Three-way concept analysis was proposed by the combination of three-way decision and formal concept analysis. Since it can provide more information than formal concept analysis, the new model has been widely used in the field of knowledge discovery. However, three-way concept analysis mainly deals with object-attribute data. If there is structure information between objects, how to remould traditional three-way concepts to describe both object-object information and object-attribute information is a topic worthy of discussion. To analyze object-object information and object-attribute information at the same time, we first propose the notion of a network formal context, and then define global network OE-concept and local network OE-concept. After that we investigate knowledge discovery based on the global network OE-concept and local network OE-concept. We further discuss the dynamic updating mechanism of global network OE-concept and local network OE-concept under the evolution of a network formal context. The corresponding algorithms are designed to update old network OE-concepts, and the time complexity is analyzed. Note that the classical formal context itself can also represent a network formal context, so in order to show the necessity of the proposed network formal context and network OE-concepts, we clarify the advantages of the network formal context and network OE-concepts from the aspects of knowledge representation, efficiency of computing network OE-concepts, dynamic updating of concepts and stability analysis of concepts. In addition, we conduct some experiments to argue that knowledge discovery of local network OE-concepts is better than that of global network OE-concepts as well as the classical OE-concepts. Meanwhile, the experimental results also show the feasibility and effectiveness of the proposed network OE-concept updating methods. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：This paper investigates two analytic methods for neural networks with time-varying delay. One is the sampled-data synchronization analysis for neural networks with time-varying delay and the other is the extended dissipative analysis for the networks with external disturbance. By constructing Lyapunov-Karasovskii functionals, looped-functionals and utilizing some mathematical techniques, a synchronization condition for neural networks with time-varying delay under the sampled-data control scheme is obtained. Improved synchronization results are proposed by adding augmented forms of functional and zero equality and applying an improved integral inequality to the previous result. And, based on the proposed criteria, the extended dissipative analysis which covers the concept of the H-infinity performance, L-2-L-infinity performance, passivity, and dissipativity is studied. Finally, two numerical examples are utilized to show the superiority and effectiveness of the proposed results. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Fake face detection is in dilemma with the rapid development of face manipulation technology. One way to improve the effectiveness of detector is to make full use of intra and inter frame information. In this paper, a novel Xception-LSTM algorithm is proposed by using our new spatiotemporal attention mechanism and convolutional long short-term memory (ConvLSTM). In the algorithm, the spatiotemporal attention mechanism, including spatial and temporal attention mechanism, is proposed to capture and enhance spatiotemporal correlations before dimension reduction of Xception. Thereafter, the ConvLSTM is introduced to consider frame structure information while modeling temporal information. The experimental results on three widely used datasets demonstrate that the proposed algorithms perform better than the state-of-the-art algorithms. In addition, the effectiveness of the spatiotemporal attention mechanism and ConvLSTM are illustrated in ablation experiments. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Although imitation learning can learn an optimal policy from expert demonstrations, it may fail to be transferred to practical environments because it is difficult to collect high-quality demonstrations for which the ultimate policy is not accurate enough and converges slowly. To solve the problem, an algorithm that utilizes Non-negative Positive-unlabeled learning (nnPU) as the probabilistic classifier to evaluate the quality of demonstrations, referred to as Non-negative Positive-unlabeled Importance Weighting Imitation Learning (PUIWIL), is proposed to increase the utilization of imperfect demonstrations and improve the performance of imitation learning. PUIWIL introduces confidence scores calculated by the nnPU classifier for expert demonstrations, which indicates the probability that the demonstration is generated by an optimal policy, and reweights all expert demonstrations according to their confidence scores. In addition, PUIWIL reconstructs the standard GAIL framework to make high-quality demonstrations have a more significant impact on imitation learning, which is called Best-in-class Imitation. The experiments demonstrate that PUIWIL improves both the performance and robustness of imitation learning from imperfect demonstrations. (C) 2022 Elsevier Inc. All rights reserved.
Silva, Rodrigo M.Gomes, Guilherme C. M.Alvim, Mario S.Goncalves, Marcos A....
24页
查看更多>>摘要：Learning to Rank (L2R) improves ranking quality but relies on the existence of manually labeled training sets, which are expensive and cumbersome to generate. Using automated labeling (e.g., clickthrough data) imposes its own challenges. Active learning (AL) can be used to gather high-quality training data by producing very informative yet small training sets. Cover, a method we have previously developed, allows for unsupervised sampling of training sets as good as those created using AL. In this paper we provide an extensive analysis of how and why Cover works. We revisit the method in a more formal way, with theorems and proofs, and provide additional empirical evidence of its practicality. We answer questions related to why Cover works so well and how its properties are related to AL methods. We show how certain characteristics of Cover's clustering step allows it to more thoroughly explore the feature space by selecting query-document pairs that are representative and diverse, allowing L2R methods to produce effective models. The main novel contribution is a detailed analysis of the method's inner workings and information-theoretic properties, allowing us to advance the understanding of L2R fundamentals through the lens of training set building. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Multi-objective model predictive control (MMPC) is an effective method to solve the problem of nonlinear systems with multiple conflicting control objectives. However, the MMPC method usually suffers from the challenge of high computational cost for the determining the optimal objective weights with the process of transforming multiple objectives into a single objective. For low computational cost, an MMPC method with gradient eigenvector algorithm (MMPC-GEA) for nonlinear systems is proposed to comprehensively deal with multiple conflicting control objectives. The proposed MMPC-GEA in the framework of MMPC is composed of a fuzzy neural network (FNN) identifier and a receding optimization algorithm. For the proposed MMPC-GEA, FNN with an adaptive learning algorithm is devised to capture the nonlinear characteristic of systems. Moreover, a gradient eigenvector algorithm (GEA) is designed to gain the optimization solution of the control objectives for nonlinear systems. Specifically, GEA can reduce the computationally demanding by avoiding the determination of the objective weights. Furthermore, the stability and control performance analysis of the MMPC-GEA scheme is provided. Finally, the effectiveness of the proposed MMPC-GEA approach is demonstrated using a numerical simulation and wastewater treatment process. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：Ride-hailing demand prediction plays an important role in ride-hailing vehicle scheduling, traffic condition control and intelligent transportation system construction. Accurate and real-time ride-hailing demand prediction is crucial for improving supply-demand imbalance, vehicle utilization and traffic conditions. However, most existing works mainly address the region-based demand prediction whereas only a few works focus on the origin-destination (OD)-based demand prediction. To address the issue, we develop several dynamic OD graphs to character the ride-hailing demand transactions between the origin and destination. We propose a novel deep learning model, referred to as the Dynamic Multi-Graph Convolutional Network with Generative Adversarial Network structure (DMGC-GAN), to investigate the challenging problem. Different from previous studies, we develop the temporal multi-graph convolutional network (TMGCN) layer with different dynamic OD graphs to capture the spatial topologies contained in the dynamic OD graphs in terms of time, and exploit GAN structure to overcome the high sparsity of OD demand. We conduct extensive numerical experiments on the real-world ride-hailing demand dataset (from Manhattan district, New York City). The results demonstrate that the model we propose performs the best against to nine baseline models. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：While recent years have witnessed the power of deep neural networks in representation learning, it is well known that their robustness is a congenital defect. Formal verification sheds some light to tackle this issue, which achieves it by a rigorous mathematical reasoning. Nevertheless, such technique still suffers from the efficiency and scalability problems. In light of this, we develop a novel solution to make a pre-analysis before performing verification. Specifically, we argue that the points near the actual decision boundary of the neural network are more likely to not satisfy robustness. As such, we focus on locating unstable points in the input set, instead of point-by-point verification. Borrowing from mutation testing, we adopt the analysis of the mutation decision boundaries to evaluate the local robustness of the inputs. Also, we design a robustness metric to guide the selection of unstable points. Then, the effective adversarial examples can be generated by perturbing these unstable points. We conduct extensive experiments on two neural network verification benchmarks, demonstrating the rationality, effectiveness and efficiency improvement of our solution. (C) 2022 Elsevier Inc. All rights reserved.
查看更多>>摘要：The continuous emergence of new malware has been a severe threat to Industrial Internet of Things (IIoT), while identifying malware through detecting malicious traffic in encrypted, drift, and imbalanced traffic streams is a challenge. This paper proposes an approach based on adaptive online analysis to accurately determine the families of malware by analyzing traffic streams which are encrypted, drift, and imbalanced. This approach is based on Improved Adaptive Random Forests (IARF), to obtain the ability of adaptive update of parameters when processing new types of malware traffic in traffic streams and being sensitive to families of malware with few samples in imbalanced traffic. We build a prototype of this approach and evaluate the performance through experiments. The experiments are based on a mixed dataset composed of data from malware-traffic-analysis.net, Lastline Inc, MCFP dataset, and CTU-13 dataset. In addition, our approach is also compared with three state-of-the-art methods. The results of the experiments show that we have obtained a 99.66% F1-score in the classification of malware families, and our classifier also performs better than the other classifiers. (C) 2022 Published by Elsevier Inc.
NSTL
Elsevier