We address robust stabilization problem for networked control systems with nonlinear uncertainties and packet losses by modelling such systems as a class of uncertain switched systems. Based on theories on switched Ly...
详细信息
ISBN:
(纸本)9781424445233
We address robust stabilization problem for networked control systems with nonlinear uncertainties and packet losses by modelling such systems as a class of uncertain switched systems. Based on theories on switched Lyapunov functions, we derive the robustly stabilizing conditions for state feedback stabilization and design packet-loss dependent controllers by solving some matrix inequalities. A numerical example and some simulations are worked out to demonstrate the effectiveness of the proposed design method.
Interactive network traffic replay is the newest method for testing and evaluation of network devices such as Firewalls, IPSes, routers, switches, etc. Currently state-checking method is used for interactive TCP traff...
详细信息
Interactive network traffic replay is the newest method for testing and evaluation of network devices such as Firewalls, IPSes, routers, switches, etc. Currently state-checking method is used for interactive TCP traffic replay. This paper proposes a new method for interactive TCP traffic replay which is based on the balance status between transmitted and received packets. By checking the balance conditions before sending out TCP packets, the method can significantly reduce the cost of state-checking and enhance the replay performance. The authors made a comparison on the differences of replay methods when introducing the balance mechanism. The efficiency of the method is also investigated and evaluated from aspects of a single TCP session, multi-session traffic, packet losses and latency. Experimental results show that the method outperforms the original state-checking method when replaying actual TCP traffics.
In some real-world classification tasks, the classifier may be trained on a data set which does not reflect the class distribution of the real data set. Such sampling bias or virtual concept drift may seriously affect...
详细信息
In some real-world classification tasks, the classifier may be trained on a data set which does not reflect the class distribution of the real data set. Such sampling bias or virtual concept drift may seriously affect the classification accuracy. Previous researches on this topic mainly concern classifiers with explicit a posteriori probabilities output. There has been a framework to adjust the original classifier using Expectation Maximization (EM) algorithm for such classifiers. The margin based classifier Support Vector Machine (SVM), has not been studied under this framework because of the lack of probabilistic output. In this paper, we discuss the probabilistic output of SVM and propose a Gaussian Mixture Model (GMM) to approximate the class conditional distribution of the margin so as to adjust the classifier using the EM framework. Experimental results on standard machine learning data sets show that the proposed algorithm can improve the classification accuracy on most of these problems. It performs especially well on those data sets with low classification accuracy.
This study relates the gait asymmetry, residual limb comfort, and energy cost during walking and identifies a compensating pattern for the trans-tibial amputees when the prostheses are misaligned. One male subject wit...
详细信息
This study relates the gait asymmetry, residual limb comfort, and energy cost during walking and identifies a compensating pattern for the trans-tibial amputees when the prostheses are misaligned. One male subject with a trans-tibial amputation volunteered for the study. The knee joint moments at the prosthetic side, the phase symmetry index, and the interface pressures were discussed under three sagittal alignment settings. The results show that the subject changes the knee joint moment, gait symmetry, and interface pressure with a misaligned prosthesis to improve his comfort and movement during walking. A high-quality liner reduces the gait sensitivity to misalignment and enhances the amputee＇s ability to compensate for misalignment. Since different people have different compensation patterns, more cases will be studied in future work.
Maximum margin clustering (MMC) is a recently proposed clustering method, which extends the theory of support vector machine to the unsupervised scenario and aims at finding the maximum margin hyperplane which separat...
详细信息
In this paper, a novel segmentation and recognition system of courtesy and legal amounts in handwritten Chinese bank checks is presented. Firstly, the system pre-recognizes the courtesy amounts by the multimodel segme...
详细信息
In this paper, a novel segmentation and recognition system of courtesy and legal amounts in handwritten Chinese bank checks is presented. Firstly, the system pre-recognizes the courtesy amounts by the multimodel segmentation and recognition algorithm. Secondly, with the guide of courtesy amount recognition candidates, the legal amount is segmented and recognized. Lastly, we fuse the recognition candidates of courtesy and legal amounts to obtain the final recognition result. By this fusion strategy, the algorithm can correct the recognition error of courtesy amounts, and the inaccurate segmentation of legal amounts. The system is validated with 1053 real bank checks. When the substitution is 0.4%, the recognition rate at the amount level can reach 66.1%.
Multiple instance learning (MIL) is a branch of machine learning that attempts to learn information from bags of instances. Many real-world applications such as localized content-based image retrieval and text categor...
详细信息
Ranking aggregation is important in data mining and information retrieval. In this paper, we proposed a semisupervised ranking aggregation method, in which the order of several item pairs are labeled as side informati...
详细信息
ISBN:
(纸本)9781595939913
Ranking aggregation is important in data mining and information retrieval. In this paper, we proposed a semisupervised ranking aggregation method, in which the order of several item pairs are labeled as side information. The core idea is to learn a ranking function based on the ordering agreement of different rankers. The ranking scores assigned by this ranking function on the labeled data are consistent with the given pairwise order constraints while the ranking scores on the unlabeled data obey the intrinsic manifold structure of the rank items. The experiment results show our method work well. Categories and Subject DescriptorsH.3.3[Information Storage and Retrieval]:Information Search and Retrieval{ Retrieval models H.3.4[Information systems Application]: systems and Software{Performance evaluation(eficiency and effectiveness).
In this paper, we present a new method of predictive deconvolution based on high order statistics, and apply it on echo cancellation. The traditional predictive deconvolution is based on the assumption that the observ...
详细信息
暂无评论