Abstract:Signed network is a kind of network including edges with the property of positive or negative sign. The positive and negative sign represent positive relationship and negative relationship, respectively. Many real complex networks have opposite relationships, especially in social, biological and information fields. Using the sign properties of edges to analysis, understand and predict the topological structures, functions and dynamic behaviors of these real networks has important theoretical significance and practical applications, such as personalized recommendation, prediction of attitudes, user analysis and clustering and so on. However, academic research has devoted little attention to signed network. This paper reviews the background, significance, research situation and recent progress of signed networks, and discusses the main problems of existing works. This study is to provide a clear and comprehensive understanding to this meaningful research area, and to benefit to the researchers from the fields of network data mining, complex network analysis, sociology and biological information.

Abstract:State explosion problem is the main obstacle of model checking. This problem is addressed in the paper from a coalgebraic point of view. By coninduction principle, the paper proves that: (1) Given any class of Kripke Structures (denoted by K), there exists a unique smallest Kripke structure (denoted by K₀) with respect to bisimilarity which describes all behaviors of the Kripke structures with no redundancy. (2) For any Kripke Structure M∈K (the state space of M may be infinite), there exists a unique concrete smallest Kripke structure K_M. Base on this idea, an algorithm is established for minimization of Kripke Structures. A naive implementation of this algorithm is developed in Ocaml. One of its applications is that instead of M, K_M can be used with a smaller state space to verify properties for M in the process of Model Checking.

Abstract:This paper introduces the notion of quantum Müller automaton (LVMA), provides the concept of quantum recognizable finite step language and the means of quantum state construction, and then proves the fact that four types of LVMA can equivalently constructed from each other. By using those equivalent relations, it establishes the algebraic and level characterizations of quantum regular infinite languages, and also explores the closed properties of these quantum infinite languages in details under some infinite regular operations in particular at the same time. Meanwhile, this study shows that the behaviors of quantum Müller automata are precisely the quantum languages definable with sentences of the monadic second-order quantum logic (LVMSO), expanding the fundamental Büchi theorem to quantum setting.

Abstract:This paper surveys the state of the art of speech emotion recognition (SER), and presents an outlook on the trend of future SER technology. First, the survey summarizes and analyzes SER in detail from five perspectives, including emotion representation models, representative emotional speech corpora, emotion-related acoustic features extraction, SER methods and applications. Then, based on the survey, the challenges faced by current SER research are concluded. This paper aims to take a deep insight into the mainstream methods and recent progress in this field, and presents detailed comparison and analysis between these methods.

Abstract:Recently, interests in learning action models have been increasing. Although non-deterministic planning has been developed for several decades, most previous studies in the field of action model learning still focus on classical and deterministic action models. This paper presents an algorithm for identifying non-deterministic actions, including effects and preconditions, in partially observable domains. It can be applied when people know nothing about a transferring system and only the action-observation sequences are given. Such scenarios are common in real-world applications. This work focuses on problems in which actions are composed of simple logical structures and features are observed under some frequency. The learning process is divided into three steps: First, compute the probability of each proposition which holds in a state. Second, extract effect schema from propositions and then extract preconditions. Third, cluster effect schema to remove redundancy. Experimental results on benchmark domains show that action model learning is still useful in non-deterministic and partial observable environments.

Abstract:Reasoning about terminological cycles of description logics is a hard problem that needs to be solved. Ordered binary decision diagram (OBDD) is a data structure suitable for dealing with large-scale problems since through the diagram Boolean functions can be represented compactly and computed efficiently. In this paper, OBDD is adopted to reason about terminological cycles of description logics. First, for terminological cycles of the description logic εL, with the help of set representation and operation, a property about the greatest simulation relationship on description graphs of terminological cycles is specified and proved. Second, by encoding description graphs of terminological cycles as Boolean functions, an OBDD-based procedure for computing the greatest simulation relationship is proposed, and based on this procedure an algorithm is presented for deciding the subsumption relationship between concepts under the greatest fix-point semantics. Third, based on OBDD, a procedure for calculating all the nodes which can reach cycle paths in a description graph is proposed, and based on this procedure an algorithm for deciding the subsumption relationship under the least fix-point semantics is also presented. Finally, the correctness and time complexity of both algorithms are proved; the computing performance of both algorithms are also demonstrated by a set of experiments. This work provides an effective approach for reasoning about terminological cycles of description logics; it also provides a new case for applying OBDD in the fields of logical reasoning.

Abstract:Forged source address and routing address prefix hijacking have caused great threats since there are no source address validation mechanisms on the current Internet. Solving the address security problem and constructing a reliable Internet environment have become a critical issue. The foundation of a trustworthy Internet is the authenticated IP addresses. Therefore, researchers have proposed many solutions from different perspectives on these problems. This paper first introduces the notion of address and the current situation of address spoofing, then gives an analysis to the meaning of the address security. The paper analyzes and compares these security solutions in three dimensions: The architecture, the mechanism and the key technical means. Their performances are also summarized and evaluated. Finally, the study provides a proposal of constructing a general experimental platform for network addresses which enables different address schemes to be deployed and experimented.

Abstract:CDN-P2P technology has recently become one of the hottest research fields. Recommending the valuable resources and improving accuracy of resource location and efficiency of content distribution for users, are CDN-P2P's challenges. This paper presents an overview of the field of CDN-P2P research in recent years from user requirements' perspective. It discusses user requirements and introduces ways of requirements acquirement. Based on CDN-P2P system over user requirements, this paper provides a comprehensive discussion on node characteristics and user similarity, relationships between nodes, nodes security and search mechanism. The difficult and hot issues about CDN-P2P system over user requirements are analyzed in depth. The future development trend and prospects for CDN-P2P system over user requirements are also discussed.

Abstract:The Internet is becoming extremely complex. Meanwhile, the network devices have been supporting more functions and services, which result in much more misconfigurations. Such misconfigurations, however, have become the main reason for network interruption as well as network anomalies. This issue has drawn many researchers' interest and attention, thus becomes a significant research topic in the field of network management. Since 2002, researchers have devoted themselves to solve configuration problems from different perspectives, and these studies greatly contribute to the development of the Internet automatic configuration. This paper firstly presents Internet automatic configuration and some configuration cases; then categorizes and evaluates Internet automatic configuration from the aspects of automatic configuration generation, configuration validation and automatic configuration realization; last but not least, summarizes the defects in the current research and then prospect the development of future research. The purpose of this paper is to provide some available information and beneficial enlightenment for researchers of this field.

Abstract:Traffic measurement in high-speed network is essential for network monitoring, management, and control. Based on the applications of network traffic measurement, this study divides the measurement into sampling methods and data stream methods. Sampling methods are partitioned into packet sampling and flow sampling, both are introduced. Data stream methods are introduced from different metrics. This study introduces in detail the common data structure and applications based on sampling and data stream methods in high-speed network. Drawbacks of different methods are analyzed and compared. The research progress of high-speed network traffic measurement technology is summarized. Finally, the limitation of recent network traffic measurement methods, the evolving trend of network traffic measurement, and some possible directions of future research are discussed.

Abstract:Internet usage has dramatically transformed from host-centric end-to-end communication to content retrieval. In order to adapt to this change, several new information/content centric networking (ICN) architectures have been proposed. One common and important feature of these architectures is to leverage built-in network caches to improve the transmission efficiency of receiver-driven content retrieval and network resource utilization. Compared with traditional Web cache and CDN cache systems, ICN cache takes on several new characteristics: Transparency, ubiquity, and fine granularity, which put new challenges on the modeling, understanding and optimization of the networked cache system. In this paper, new features of ICN as well as their ensuing challenges are introduced, mechanisms to optimize the performance of cache network are analyzed and compared from various angles, and analytical models for the cache system are also presented. The study concludes with a discussion on some key issues and future research directions.