In this paper, we report our recent research activities under MICIE, a European project funded under Framework-7 Programme, in which a SVM-based behavior modeling and learning algorithm is described. The proposed algo...
详细信息
ISBN:
(纸本)9783642040900
In this paper, we report our recent research activities under MICIE, a European project funded under Framework-7 Programme, in which a SVM-based behavior modeling and learning algorithm is described. The proposed algorithm further exploits the adapted learning capability in SVM by using statistics analysis and K-S test verification to introduce an automated parameter control mechanism, and hence the SVM learning and detection can be made adaptive to the statistics of the input data. Experiments on telecommunication network data sets support that the proposed algorithm is able to detect undesired events effectively, presenting a good potential for development of computer-aided monitoring software tools for protection of critical infrastructures.
After a first phase of great activity in the field of multi-agent systems, researchers seemed to loose interest in the paradigm, mainly due to the lack of scenarios where the highly distributed nature of these systems...
详细信息
ISBN:
(纸本)9783642004865
After a first phase of great activity in the field of multi-agent systems, researchers seemed to loose interest in the paradigm, mainly due to the lack of scenarios where the highly distributed nature of these systems could be appropriate. However, recent computing models such as ubiquitous computing and ambient intelligence have introduced the need for this type of highly distributed, autonomous and asynchronous computing mechanisms. The agent paradigm can play an important role and can suit the needs of many applications in these scenarios. In this paper we argue that the main obstacle for the practical application of multi-agent systems is the lack of appropriate security mechanisms. Moreover, we show that as a result of recent advances in security technologies, it is now possible to solve the most important security problems of agent-based systems.
This research aims at developing "Humanoid Robots" that can carry out intellectual conversation with human beings. The first step of our research is to recognize human emotions by a computer using neural net...
详细信息
ISBN:
(纸本)9783642032011
This research aims at developing "Humanoid Robots" that can carry out intellectual conversation with human beings. The first step of our research is to recognize human emotions by a computer using neural network. In this paper all six universally recognized principal emotions namely angry, disgust, fear, happy, sad and surprise along with neutral one are recognized. Various neural networks such as Support Vector Machine (SVM), Multi layer Perceptron (MLP), Principal Component Analysis (PCA), and Generalized Feed Forward Neural Network (GFFNN) are employed and their performance is compared. 100% recognition accuracy is achieved on training data set (seen examples) and test data set (unseen examples).
The paper presents initial research on the system for remote guidance of the blind. The concept is based on the idea that a blind pedestrian can be aided by spoken instructions from an operator who receives a video st...
详细信息
ISBN:
(纸本)9783642032011
The paper presents initial research on the system for remote guidance of the blind. The concept is based on the idea that a blind pedestrian can be aided by spoken instructions from an operator who receives a video stream from a camera and GPS readouts from a sensor carried by a visually impaired user. An early prototype utilizing two laptop PCs and a wireless internet connection is used in indoor orientation and mobility trials, which aim to measure the potential usefulness of the system and discover possible problems with user-operator communication or device design. A second prototype is also constructed using a smaller subnotebook Flybook computer, a GPS sensor and GSM modem with HSDPA technology, and tested in outdoor environments. Test results show a quantitative performance increase when travelling with a remote guide: 15-50% speed increase and halved times of navigational tasks in indoor trials, as well as a significant decrease in the number of missteps and accidental obstacle collisions in outdoor trials. A large part of the success is the supportive feedback from the blind testers, who point out the engendered feeling of safety when assisted.
Rough sets offer an effective approach of managing uncertainties and can be employed for tasks such as data dependency analysis, feature identification, dimensionality reduction, and pattern classification. As these t...
详细信息
ISBN:
(纸本)9783540896180
Rough sets offer an effective approach of managing uncertainties and can be employed for tasks such as data dependency analysis, feature identification, dimensionality reduction, and pattern classification. As these tasks are common in many medical applications it is only natural that rough sets, despite their relative 'youth' compared to other techniques, provide a suitable method in such applications. In this paper, we provide a short summary on the use of rough sets in the medical informatics domain, focussing on applications of medical image segmentation, pattern classification and computer assisted medical decision making.
This paper presents an alternative voice conversion technique using support vector machine (SVM)-regression as a tool which converts a source speaker's voice to specific standard target speaker. The main objective...
详细信息
ISBN:
(纸本)9783540896180
This paper presents an alternative voice conversion technique using support vector machine (SVM)-regression as a tool which converts a source speaker's voice to specific standard target speaker. The main objective of the work is to capture a nonlinear mapping function between the parameters for the acoustic features of the two speakers. Line spectral frequencies (LSFs) have been used as features to represent the vocal tract characteristics. We use kernel induced feature space with radial basis function network (RBFN) type SVM that uses gaussian kernel. The intonation characteristics (pitch contour) is modified using the baseline technique, i.e. gaussian normalization. The transformed LSFs along with the modified pitch contour are used to synthesize the speech signal for the desired target speaker. The target speaker's speech signal is synthesized and evaluated using both the subjective and the listening tests. The results signify that the proposed model improves the voice conversion performance in terms of capturing the speaker's identity as compared to our previous approach. In the previous approach we used feed forward neural network (FFNN) based model for vocal tract modification and codebook based method for pitch contour modification. However, the performance of the proposed system can further be improved by suitably modifying various user-defined parameters used in regression analysis and using more training LSF vectors in the training stage.
The World softcomputing (WSC) conference is an annual international online conference on applied and theoretical softcomputing technology. The topics covered by the conference range from applied to theoretical aspec...
ISBN:
(数字)9783540896197
ISBN:
(纸本)9783540896180
The World softcomputing (WSC) conference is an annual international online conference on applied and theoretical softcomputing technology. The topics covered by the conference range from applied to theoretical aspects of fuzzy, neuro-fuzzy and rough sets over to neural networks to single and multi-objective optimisation. Contributions about particle swarm optimisation, gene expression programming, clustering, classi cation, support vector machines, quantum evolution and agent systems have also been received. One whole session was devoted to softcomputing techniques in computer graphics, imaging, vision and signal processing.
Evolutionary agents are flexible, agile, capable of learning, and appropriate for problems with changing conditions or where the correct solution cannot be known in advance. Evolutionary Multi-Agent systems, therefore...
详细信息
ISBN:
(纸本)9783642004865
Evolutionary agents are flexible, agile, capable of learning, and appropriate for problems with changing conditions or where the correct solution cannot be known in advance. Evolutionary Multi-Agent systems, therefore, consist of populations of agents that learn through interactions with the environment and with other agents and which are periodically subject to evolutionary processes. In this paper we present a JADE-based programming framework for creating evolutionary multi-agent systems with the aim of providing all the necessary infrastructure for developing multi-agent systems of this type. Through its graphical interface, the framework allows to easily configure the parameters of the multi-agent system, to hold complete control over its execution, and to collect performance data. This way the development of an evolutionary MAS is simplified and only little pieces of code have to be written in order to apply the framework to a particular problem. Along this paper, the features of the framework are described and its capabilities and usage are illustrated through its application to the tic-tac-toe problem.
Nowadays new intelligent techniques have been used to improve the intrusion detection process in distributed environments. This paper presents an approach to define an ontology model for representing intrusion detecti...
详细信息
ISBN:
(纸本)9783642040900
Nowadays new intelligent techniques have been used to improve the intrusion detection process in distributed environments. This paper presents an approach to define an ontology model for representing intrusion detection and prevention events as well as a hybrid intelligent system based on clustering and Artificial Neuronal Networks for classification and pattern recognition. We have specified attacks signatures, reaction rules, asserts, axioms using Ontology Web Language with Description Logic (OWL-DL) with event communication and correlation integrated on Multi-Agent Systems. incorporating supervised and unsupervised models and generating intelligent reasoning.
暂无评论