A Convolutional Neural Network Framework for Accurate Skin Cancer Detection

Neural Processing Letters - Skin diseases have become a challenge in medical diagnosis due to visual similarities. Although melanoma is the best-known type of skin cancer, there are other...     

针对网络服务提供商的一种流量分类和应用架构

数据流量识别技术在应对网络病毒、网络入侵检测和网络管理等方面部有着重要的应用。当前主要的网络流量识别技术主要是DPI（Deep Packet Inspection）和DFI（Deep Flow Inspection）,两种技术都各自有优点和缺点。本史提出了一种网络架构,将DPI和DFI技术相结合,并通过数据挖掘技术来辅助发现分类规则,应用FPGA的可编程性和并发性来加速识别和处理过程。     

基于网络演算的流量整形模型

流量整形对QoS(quality of service)控制有着重要的影响.总结并优化了最近几年发展起来的能够深刻透视计算机网络业务流控制问题的网络演算技术,采用网络演算技术建立了包括贪婪无损失整形器、无缓冲区整形器及固定缓冲区长度整形器的一般性模型,获得了3种整形器的输入/输出特性以及整形器可提高QoS控制机制的效率等结果.     

An Architectural Modelfor Intelligent Network Management

1 IntroductionThe explosion in size and complexity of today's computer network means that NetworkManagement (NM) is a critical factor. The ever growing demand for distributed applicationscoupled with increasing corporate globalization often leads to users requiring constant network availability as the network becomes the primary strategic resource. The currellt NMbecomes increasingly inadequate, since the networks to be managed are heterogeneous anddynamic, and manapemellt functions, such as…     

一种检测网络流量异常和网络攻击的算法

尽管网络流量会出现异常,但大部分时间里流量变化是有规律的。如果网络流量出现异常,那么对整个网络具有较大的危害性,因此网管人员必须及时发现流量是否异常。在以往的流量检测算法中聚焦的焦点只是如何区分流量是正常还是异常,并未讨论此时是否存在攻击行为,这些是算法的不足之处。为了改进现有算法的不足之处,作者在查阅了有关资料和文献之后,提出了一种新的流量检测算法,该算法使用了相似性计算法算法,MMTD算法和粗糙集中的决策系统,将这三种算法在流量检测中进行应用是本文的创新点。本文解决问题的思路是首先使用MMTD算法对当前的流量是否异常做出判断,如果出现异常,则使用相似性计算算法将此时的流量与预警值进行比较,并且判断此时是否存在攻击行为,在文章的最后使用粗糙集中的决策系统对流量异常的原因做出决策。     

一种基于多Agent服务的网络体系结构

本文从 Ａ Ｔ Ｍ 模型出发,介绍了一种基于多 Ａｇｅｎｔ服务的网络体系结构。本结构在硬软件方面提供通信服务、链接许可控制及信令服务等功能;支持以用户为中心的 Ｑｏ Ｓ 规程,允许多 Ａｇｅｎｔ 满足终端用户的各种服务请求。本结构之原型模拟系统经运行后,初步实现了内部带宽存储等机制,有效地满足了用户的 Ｑｏ Ｓ需求。     

一种支持流量工程的MPLS网络改进模型

本文提出了一种根据业务特性对网络应用进行分类的方法，设计了一种支持流量工程的MPLS网络改进模型，给出了该模型的路由机制、资源分配、服务质量保证和流量工程策略等机制，最后分析了该网络模型的利弊。     

An Adaptive Policy-Based Framework for Network Services Management

This paper presents a framework for specifying policies for the management of network services. Although policy-based management has been the subject of considerable research, proposed solutions are often restricted to condition-action rules, where conditions are matched against incoming traffic flows. This results in static policy configurations where manual intervention is required to cater for configuration changes and to enable policy deployment. The framework presented in this paper supports automated policy deployment and flexible event triggers to permit dynamic policy configuration. While current research focuses mostly on rules for low-level device configuration, significant challenges remain to be addressed in order to:a) provide policy specification and adaptation across different abstraction layers; and, b) provide tools and services for the engineering of policy-driven systems. In particular, this paper focuses on solutions for dynamic adaptation of policies in response to changes within the managed environment. Policy adaptation includes both dynamically changing policy parameters and reconfiguring the policy objects. Access control for network services is also discussed.     

An Integrated Framework for Implementing Quality of Network Concepts

This paper presents an integrated framework for describing and implementing quality of network (QoN) concepts. It focuses on identifying a systems engineering framework for better understanding how to specify and implement QoN concepts. We demonstrate, by means of several examples, that the quality of service interfaces between the end user and the underlying network are very important. Finally, we provide insights as to future development efforts needed to realize the goal of high quality networks.     

一个嵌入式网络流量识别系统的设计与实现

文章介绍一个在嵌入式环境下的网络流量识别系统设计并实现,该系统具有可扩展的特征。采用流量特征配置、Linux内核模块等技术灵活扩展系统识别能力。本文剖析了流量识别、流量分类和系统扩展接口等高效的实现方法。通过测试验证了谊系统能正确有效的识别企业内部网的网络流量。     

QoS体系中流量工程与网络规划的功能框架

概述了Internet网络的发展趋势，分析了QoS提供体系的研究现状，指出了现有研究的不足，然后，将网络规划引入QoS提供体系架，把网络规划和流量工程结合起来，拓展了原有的QoS提供体系框架，最后，提出了网络规划和流量工程综合框架的系统流程，并且严区分了网络规划，流量工程和网络优化的作用范围。     

OAuthkeeper: An Authorization Framework for Software Defined Network

Implementing REST API for SDN is quite challenging compared to conventional web services. First, the state transfers in SDN are more complex among network devices, controllers, and applications. Second, SDN provides more granular resources in both the controller and the network device itself. Those challenges require SDN to have a proper REST API security definition, which is currently not available in most of the SDN controllers. In this paper, we propose and implement a REST API security module for SDN controller based on OAuth 2.0. We answer the SDN REST API security challenges by presenting novel access control parameters to cope with the granular resources introduced by SDN. Our prototype maintains the best trade-off between performance and safety by generating a maximum value of 15% overhead during our benchmark. It also offers a customizable and flexible access control for the network in various use cases.     

An Extensible Heterogeneous Network Embedding Framework for Knowledge Tracing

Knowledge tracing is of great significance for providing better personalized learning guidance and has thus attracted extensive research attention in recent years. The task of knowledge tracing is to model students'' learning process on the basis of historical exercise records and trace students'' knowledge proficiency, thereby predicting students'' performance on future exercises or recommending exercises for better proficiency. Existing methods focus on either the skill level or the exercise level, ignoring the relationships among exercises and Knowledge Components (KCs). The classical single-factor models include the Deep Knowledge Tracing (DKT) {model} and the Dynamic Key-Value Memory Network (DKVMN) model. Although a few models, such as the Bayesian Knowledge Tracing (BKT) model and the Knowledge Proficiency Tracing (KPT) model, utilize the Q-matrix to improve model performance, most of them ignore the interaction among KCs, not to mention models that do not use the Q-matrix. Inspired by the recent success of network embedding, this paper presents a heterogeneous network embedding framework for knowledge tracing called HNEKT that takes both exercises and KCs into account. To adapt to the application of knowledge tracing, this paper also proposes several meta-paths to generate meaningful node sequences for network embedding. Besides, it explores other side information as well to improve the extensibility and effectiveness of the proposed model. Extensive experiments on three real-world datasets demonstrate the effectiveness of the HNEKT model.     

一个移动Agent在IP网络流量监控中的应用

介绍了当前IP网络流量监控的主要方法，提出了一种运用移动代理技术来实现对IP网络流量有效监控的方法，阐述了移动代理环境，运用移动代理进行流量监控的过程，以及移动Agent和SNMPAgent的交互操作。     

一种广电运营商HFC网络业务接纳控制架构

在广电运营商的业务提供系统中,HFC网络业务请求的接纳控制是一项关键技术。首先分析了HFC网络业务请求接纳控制面临的问题,在此基础上提出了一种接纳控制的统一架构。该架构将针对业务请求的接纳控制策略分为接纳判定策略和频点选择策略,其中,接纳判定策略用于对业务请求是否允许接纳进行控制,频点选择策略用于对业务请求接纳到哪个频点进行控制。在该架构中这两种策略采用一个统一的流程进行调用。采用该架构可以支持已有业务请求接纳控制策略的灵活调整和新业务请求接纳控制策略的灵活扩展。     

基于选择性集成策略的嵌入式网络流特征选择

机器学习在网络流量分类中存在特征选择度量指标单一、类别不平衡和概念漂移等问题,使得模型复杂度提高、泛化能力下降.该文提出基于选择性集成策略的嵌入式特征选择方法,根据选择性集成策略选取部分特征选择器集成,再改进序列前向搜索和封装器组合方法二次搜索最优特征子集.实验结果表明该算法在保证分类效果的同时有效降低了特征子集复杂度,从而达到了分类效果、效率和稳定性的最优平衡.     

An Architecture for Network Congestion Control and Charging of Non-cooperative Traffic

This paper proposes an architecture capable of reducing network congestion caused by the intense use of non-cooperative traffic. A charging scheme is imposed to all traffic that is carried over the UDP protocol (non-cooperative) given its intrinsic priority over TCP. Prices are calculated according to the degree of starvation undergone by cooperative TCP flows. When TCP flows experience a low performance, charges are high for non-cooperative flows and so the architecture tends to block new incoming UDP traffic. Knowledge about flows status is obtained through the use of flow protocol technology. Resources are reserved using firewall rules and custom-queueing mechanisms. An implementation of the architecture is made and tests show the effectiveness of our proposal in a real network scenario.     

网络繁心程度的评价方法研究与实现

提出了一种基于概率论观点的归一量化算法,该算法通过将网络繁忙程度归一量化,从而完成对当前网络性能的评价,并将评价结果用曲线记录下来,用随机变量的数字特征描述该曲线。这种有记忆怕归一量化标准能够体现网络的实际情况,具有简单、全面的特点。在程序设计中,还采用了多线程的设计思想,使系统能快速地响应Windows消息。     

一种网络流量监控算法设计

网络流量监控是实现网络性能分析和监控的重要手段,文章阐述了采用排队论建立网络流量分析的基本模型,得出网络的流量预测方式和稳态拥塞率求解公式,结合常用的网络流量监控参数,合理的实现对网络流量的估算和监控过程。