Showing papers by "Lakhmi C. Jain published in 2003"

Recent advances in intelligent paradigms and applications

Abstract: 1. Intelligent systems: architectures and perspectives.- 2. Hybrid architecture for autonomous robots, based on representations, perception and intelligent control.- 3. An intuitionistic fuzzy set based approach to intelligent data analysis: an application to medical diagnosis.- 4. A fuzzy inference methodology based on the fuzzification of set inclusion.- 5. A fuzzy approach to job-shop scheduling problem based on imprecise processing times.- 6. On efficient representation of expert knowledge by fuzzy logic.- 7. Discovering efficient learning rules for feedforward neural networks using genetic programming.- 8. Neuro-fuzzy methods for modeling and identification.- 9. Constrained two dimensional bin packing using a genetic algorithm.- 10. Sequential and distributed evolutionary algorithms for combinatorial optimization problems.- 11. Embodied emotional agent in intelligent training system.- 12. Optimizing intelligent agent's constraint satisfaction with neural networks.

A framework for interfacing BDI agents to a real-time simulated environment

Abstract: This paper describes a framework for interfacing BDI-based Agents within a real-time simulation environment. The test bed environment presented is based on a commercial game called Unreal Tournament and the Agents are developed using JACK, a commercial BDI Agent development language. The interface extends previous work done by the Gamebots and Javabots projects and effectively links an Unreal Tournament character in the game with the BDI reasoning structures of the agent that controls the character. The structure of a simple demonstration agent that uses the interface will also be described.

A novel VQ-based watermarking scheme with genetic codebook partition

Abstract: A novel watermarking scheme based on vector quantisation (VQ) for digital still images is presented. This scheme begins with the procedure of partitioning the original codebook into two sub-codebooks. To achieve this, two strategies are proposed. The first one requires no complex algorithm and gives the users full freedom for partitioning. The second one is a genetic codebook partition procedure, which has the ability to improve the perfonmance of tile proposed watermarking scheme. After that, the information of codebook partition is served as a secret key and is used in the proposed watermarking scheme. In the embedding procedure, according to the watermark bit to be embedded a sub-codebook is chosen. The traditional VQ nearest codeword search is then performed to obtain the nearest codeword for the input vector. In the extracting procedure, the traditional VQ table lookup procedure is executed. With the same secret key, the hidden watermark bit can be determined by examining which same-codebook the corresponding codeword is belonging to. It is denaonstated that the proposed watermarking scheme provides better quality in watermarked images, stronger robustness under some common attacks, faster encoding time, and effective methods for partitioning codebooks.

A concurrent fuzzy-neural network approach for decision support systems

Abstract: Decision-making is a process of choosing among alternative courses of action for solving complicated problems where multi-criteria objectives are involved. The past few years have witnessed a growing recognition of Soft Computing technologies that underlie the conception, design and utilization of intelligent systems. Several works have been done where engineers and scientists have applied intelligent techniques and heuristics to obtain optimal decisions from imprecise information. In this paper, we present a concurrent fuzzy-neural network approach combining unsupervised and supervised learning techniques to develop the Tactical Air Combat Decision Support System (TACDSS). Experiment results clearly demonstrate the efficiency of the proposed technique.

TACDSS: Adaptation Using a Hybrid Neuro-Fuzzy System

Abstract: Normally an intelligent decision support system is build to solve complex problems involving multi-criteria decisions. The knowledgebase is the vital part of the decision support system containing the knowledge or data that is used for decision-making. Several works have been done where engineers and scientists have applied intelligent techniques and heuristics to obtain optimal decisions from imprecise information. In this paper, we present a hybrid neuro-fuzzy technique for the adaptive learning of Takagi-Sugeno type fuzzy if-then rules for the Tactical Air Combat Decision Support System (TACDSS). Experiment results clearly demonstrate the efficiency of the proposed technique. Some simulation results demonstrating the difficulties to decide the optimal number and shape of the membership functions are also provided.

Fair multi-path selection for real-time video transmission in ad-hoc networks using artificial intelligence

Abstract: The transmission of video data over dynamically connected ad-hoc networks is challenging. Strict minimum delay and low bit error rate are the main requirements for real-time video transmission. Layer Coding and Multi-path Description Coding are the two most popular source coding solutions for this task. MDC and LC both require a multi-path protocol to achieve high performance in video playback. Naturally a suitable multi-path selection would determine a set of shortest paths between source and destination to satisfy the real-time requirements. However the high contention for the shortest path would lead to ineffective use in network resources. In this paper a multi-path selection scheme is introduced using appropriate feedback information, fuzzy logic and depth first search. The aim of path selection is to achieve the optimal balance between effective use of network resources and real-time video requirements.

Expanded Neural Networks in System Identification

Abstract: The neural networks are recognized to possess the fault tolerance and learning capability. The neural networks are also used in the identification of nonlinear systems. However in the system identification it is important to whiten a color noise using the noise model. In this paper we propose an expanded neural network in which a noise model is incorporated into the output layer of the neural network. We have developed the learning algorithm converged more quickly than a classical back-propagation algorithm. The proposed algorithm estimates the parameter of the expanded neural network using the least-squares method, and estimates threshold by the fundamental error back-propagation method.

Fuzzy tracking expert system for enhancing situation awareness

Abstract: This paper describes a Fuzzy Tracking Expert System for enhancing situation awareness and providing decision support for future land defence applications. The system has two modules; a Fuzzy Identification Module and a Fuzzy Tracking Module, which are both implemented using Java programming language. The Fuzzy Identification Module monitors the movement of targets in the battlefield and classifies each target as friend or enemy depending on the distances between the target to existing enemy routes and an enemy base. The Fuzzy Tracking Module creates new tracks and builds up existing tracks by correlating observations to existing tracks based on target's heading, velocity and the reliability of the input data.

Knowledge engineering in an intelligent environment

Abstract: The concept of knowledge engineering, starting with its deep association with information management, still carries multiple, even conflicting interpretations. The most popular one being a structured field that encompasses processes and techniques for knowledge discovery, indexing, organization, and fusion. Where the classical approach to knowledge engineering and management tends to rely on techniques like concept maps, hypermedia and object-oriented databases, computational intelligence techniques for core knowledge engineering activities like knowledge discovery, organization, and knowledge fusion are rapidly gaining popularity. In the evolved scenario, knowledge engineering may be interpreted as a field that deals with acquisition, storage and application of knowledge for a range of knowledge intensive tasks – whether it is decision support, learning or research support. This special issue on ‘Knowledge Engineering in an Intelligent Environment’ is an attempt to present some of the latest theoretical and application developments in the field of knowledge engineering. This special issue comprises of four papers on different aspects of knowledge management and is organized as follows. In the first paper Jermol et al. present a virtual enterprise model used in networking international expert teams from academia and business in the area of data mining and decision support. The knowledge management aspects of business intelligence as implemented in the virtual enterprise model are analyzed in terms of appropriate business organizational and management models. Further, construction of a knowledge map of the available tools, expertise and collaborative work procedures, cognitive authority in collaborative work management, as well as the network intelligence aspect of the virtual enterprise endeavor are discussed. Authors made use of a European virtual enterprise as a case study to illustrate some of the lessons learned. Messina et al. in the second paper discuss a rigorous approach to engineer the knowledge within intelligent controllers. The key to real-time intelligent control lies in the knowledge models that the system contains. Authors identified three main classes of knowledge namely parametric, geometric/iconic, and symbolic and examples are illustrated. Each of these classes provides unique perspectives and advantages for the planning of behaviors by the intelligent system. Since the early eighties, there has been a gradual shift in the focus of development of knowledge based systems away from the rapid prototyping techniques that had previously prevailed, toward more structured methodologies, including model based reasoning and modeling of knowledge domains. The default standard for the development of these systems has become the CommonKADS methodology. In the third paper