As the Internet is shifting towards a reliable QoS-aware network, accurately synchronized clocks distributed in the Internet become more significant. Network Time Protocol (NTP) is broadly deployed in the Internet for...
详细信息
ISBN:
(纸本)0769514502;0769514510
As the Internet is shifting towards a reliable QoS-aware network, accurately synchronized clocks distributed in the Internet become more significant. Network Time Protocol (NTP) is broadly deployed in the Internet for clock synchronization among distributed hosts, but it is weak in asymetric paths, i.e., it cannot accurately estimate the clock offset between two hosts when the forward and backward paths between them have different one-way delays. In this paper, we focus on estimating the offset and the skew of a clock from one-way delay measurement between two hosts, and propose an idea for improvement of such estimations, which reduces estimation errors in case that the forward and backward paths have different bandwidths that is one of main factors of the asymmetric delays.
In the Internet, a statistical perspective of global traffic flows has been considered as an important key to network management. Nonetheless, it is expensive or sometime difficult to measure statistics of each flow s...
详细信息
ISBN:
(纸本)0769514472
In the Internet, a statistical perspective of global traffic flows has been considered as an important key to network management. Nonetheless, it is expensive or sometime difficult to measure statistics of each flow separately. Therefore, it is of practical importance to infer unobservable statistical characteristics of individual flows front characteristics of the aggregated-flows that are easily measured at some links (router interfaces) in the network. In this paper, we propose a new approach to such inference problems, and provide some examples of inferring unobservable arrival rates of packets on each flow front measurement of the aggregated-flows. Our method is applicable to cases not covered by the existing methods for the OD traffic matrix inference. We also show simulation results, which indicate potential of our approach.
For two-player games of perfect information such as Chess, we introduce "uniqueness" properties to describe game positions in which a winning strategy (should one exist) is forced to be unique. Depending on ...
详细信息
In this paper we present a simulation environment for the study of hierarchical job scheduling on distributed systems. The environment provides a multi-level mechanism to simulate various types of jobs. An execution m...
详细信息
Instance selection and feature selection are two orthogonal methods for reducing the amount and complexity of data. Feature selection aims at the reduction of redundant features in a dataset whereas instance selection...
详细信息
ISBN:
(纸本)9781581135671
Instance selection and feature selection are two orthogonal methods for reducing the amount and complexity of data. Feature selection aims at the reduction of redundant features in a dataset whereas instance selection aims at the reduction of the number of instances. So far, these two methods have mostly been considered in isolation. In this paper, we present a new algorithm, which we call FIS (Feature and Instance Selection) that targets both problems simultaneously in the context of text classification. Our experiments on the Reuters and 20-Newsgroups datasets show that FIS considerably reduces both the number of features and the number of instances. The accuracy of a range of classifiers including Naïve Bayes, TAN and LB considerably improves when using the FIS preprocessed datasets, matching and exceeding that of Support Vector Machines, which is currently considered to be one of the best text classification methods. In all cases the results are much better compared to Mutual Information based feature selection. The training and classification speed of all classifiers is also greatly improved.
In many applications the image registration task turns out to be a fundamental prerequisite for any further processing and analysis. In the proposed paper a new efficient technique is presented, appropriate for regist...
详细信息
In many applications the image registration task turns out to be a fundamental prerequisite for any further processing and analysis. In the proposed paper a new efficient technique is presented, appropriate for registering images which are translated and rotated versions of a reference image. The technique consists of two main parts: an efficiently implemented prewhitening part, and an iterative part which yields the unknown displacements after a few iterations. The cross-correlation operations involved in the iterative part are performed efficiently via a new scheme based on a proper partitioning of the images and the use of the FFT. Note that this efficient scheme is also applicable to the conventional spatial cross-correlation method. The new registration technique exhibits a superior performance as compared to the conventional cross-correlation method.
Herein, we propose a new Markov random field (MRF) image segmentation mode! which aims at combining color and texture features. The model has a multi-layer structure: Each feature has its own layer, called feature lay...
详细信息
Herein, we propose a new Markov random field (MRF) image segmentation model which aims at combining color and texture features. The model has a multi-layer structure: Each feature has its own layer, called feature lay...
详细信息
We study the problem of allocating optical bandwidth to sets of communication requests in all–optical networks that utilize Wavelength Division Multiplexing (WDM). WDM technology establishes communication between pai...
详细信息
暂无评论