One of the open problems in neural network research is how to automatically determine network architectures for given applications. In this brief, we propose a simple and efficient approach to automatically determine the number of hidden nodes in generalized single-hidden-layer feedforward networks (SLFNs) which need not be neural alike. This approach referred to as error minimized extreme learning machine (EM-ELM) can add random hidden nodes to SLFNs one by one or group by group (with varying group size). During the growth of the networks, the output weights are updated incrementally. The convergence of this approach is proved in this brief as well. Simulation results demonstrate and verify that our new approach is much faster than other sequential/incremental/growing algorithms with good generalization performance.

Abstract A collaborative virtual environment (CVE) allows geographically dispersed users to interact with each other and objects in a common virtual environment via network connections. One of the successful applications of CVE is multiplayer on‐line role‐playing game. To support massive interactions among virtual entities in a large‐scale CVE and maintain consistent status of the interaction among users with the constraint of limited network bandwidth, an efficient collaborative interaction management method is required. In this paper, we propose a behaviour‐based interaction management framework for supporting multiplayer role‐playing CVE applications. It incorporates a two‐tiered architecture which includes high‐level role behaviour‐based interaction management and low‐level message routing. In the high level, interaction management is achieved by enabling interactions based on collaborative behaviour definitions. In the low level, message routing controls interactions according to the run‐time status of the interactive entities. Collaborative Behaviour Description Language is designed as a scripting interface for application developers to define collaborative behaviours of interactive entities and simulation logics/game rules in a CVE. We demonstrate and evaluate the performance of the proposed framework through a prototype system and simulations. Copyright © 2006 John Wiley & Sons, Ltd.

This paper describes a large-scale collaborative virtual environment architecture that has the capability of scaling across multiple geographically dispersed resources. The architecture consists of distributed mobile agents working cooperatively in supporting and managing the virtual environments. The mobile agents' tasks include managing persistency and consistency of the virtual worlds, maintaining reliability and efficiency of user interactions, ensuring security and integrity of data and systems, and so on. The mobile agents are autonomous and has the ability of migrating among hosts to maximize resource utilizations. Grid technologies and concepts are introduced into the architecture. These improvements allow the mobile agents to execute and communicate securely in multiple administrative domains. Grid-based scheduling components and polices are integrated to provide intelligent resource optimizations. Furthermore, a better load-balancing can be achieved by utilizing additional or more accurate information like data-user proximity and hosts' workload invariance etc. The result will be a more scalable and robust mobile agent architecture.

In order to achieve entity evolvement, enrich socialization of virtual community and optimize computer processing in the collaborative virtual environment (CVE), efficient collaborative interaction management (CIM) is a key to handle these issues. In this paper, we focus on functional extensibility in CVE, and give the solution by using the conceptual idea of our proposed CIM approach, called task-oriented interaction management. This approach has four important components, which are scene interaction manager, virtual entity interaction manager, interaction task agent, and scene role & rules. They have certain capabilities, like data distribution, load-balance, tasks migration, ownership transfer, entities' states consistency maintenance, database persistency management and etc. The motivation of this research is to develop an extensible and evolving CVE with extensible functionalities for creating large-scale collaborative interaction experience to users in a large-scale interconnected 3D virtual environment. Details of the conceptual framework of the task-oriented interaction management will be presented in this paper.

Collaborative virtual environment (CVE) is a promising technology which provides people a 3D shared virtual world over the networks, where participants can interact with each other more naturally. However, the scalability of existing CVE system is limited due to the constraints in processing power and network speed of each participating host. On the other hand, mobile agent is a new paradigm for the network-centric programming which can implement distributed applications in an open and dynamically changing environment. In this paper, a mobile agent-based architecture for large-scale CVE (LCVE) is proposed to support a large number of concurrent participants in a VE with a large amount of evolving virtual entities. The LCVE software system is made up of mobile agents. And each mobile agent will be responsible for an independent task for the LCVE. With the mobility and autonomy of mobile agents, LCVE software system will be extensible; the workload in the LCVE will be balanced dynamically; the management of LCVE contents will be autonomous at agent level.

The capabilities of Virtual Reality Modeling Language (VRML)/Extensible 3D (X3D) permit building collaborative virtual environments (CVE) using the Internet and the World Wide Web. However, they do not provide a multi-user collaborative interaction management mechanism for CVE applications except the External Authoring Interface that allows communication between external program and VRML browser. The authors of such applications must manage the interactions among the interactive entities by themselves through capturing user interaction with particular virtual entities and propagating them to the affected users. This application-specific method leads to inflexibility in reusing collaborative role definitions and adapting various collaborative interaction management methods. In this paper, we propose a Collaborative Behavior Description Language(CDBL) to address the identified issues. The syntax and sample applications of CBDL will be presented in details. Our work shows that CBDL enables the reuse of collaborative role definitions and scene descriptions to construct flexible CVE with adaptive underlying collaborative interaction management methods.

In this paper, we propose a mechanism for managing the collaborative interactions among the distributed users in our developed SmartCU3D, an Internet Collaborative Virtual Environment (CVE) system. In this mechanism, the message routing in the system becomes adaptive to the application and the user’s runtime interaction. It’s achieved by giving the ObjectOriented style collaborative behavior description along with the 3D environment definition for every individual CVE application. The collaborative behavior description defines the functional roles selected by the users in the application, and the behaviors for every functional role. The interaction messages for maintaining the virtual world’s consistence are generated according to the behavior definition of the interactive entity’s functional role. And the interaction message will be routed among the affected interactive entities in the Internet CVE. The motivation of this research is to achieve a framework to support a flexible and adaptive interaction management mechanism for developing different CVE applications.

To support real-time natural interaction in the Collaborative Virtual Environment (CVE) with limited network bandwidth and computer processing power, the development of an efficient interaction management mechanism is a key issue. In this paper, we propose a behavior based interaction management mechanism for managing the collaborative interactions among the distributed users in our developed SmartCU3D, an Internet CVE system. With this mechanism, message routing in the system becomes adaptive to the application and the users' runtime interaction. It is achieved by giving the Object-Oriented style collaborative behavior description along with the 3D environment definition for every individual CVE application. The motivation of this research is to develop a framework to support a flexible and adaptive interaction management mechanism for developing different CVE applications.

This paper addresses the problem of video server selection in a distributed video-on-demand system. The objective of the problem is to balance the load among video servers and to minimize the delay for the video requests to be served. In order to achieve this objective, a near optimal assignment algorithm is proposed for video server selection. Empirical study shows that our proposed algorithm is able to achieve good load balancing performance as compared to some landmark algorithms.

Existing approaches of tele-operation of underwater robots using a video camera are very inefficient due to poor visibility around the worksite and limited camera field of view. This paper describes research work which attempts to overcome the problem by introducing a virtual telepresence operation approach. This virtual telepresence interface takes the robot's position and orientation data from a sonar navigation system, and generates 3D synthetic images of the worksite based on its CAD model using virtual reality technology. It provides the robot operators with a full perception of its spatial location, flexible options of viewpoints and functions for tele-operation of underwater robots. The interface can also be used for training of the robot operators. The main conclusion is that virtual telepresence incorporated with a robot safety domain concept not only can improve the efficiency of operation of underwater robots, but also forms the basis for supervisory control and fully automated control of the robots.