Showing papers presented at "Intelligent Systems Design and Applications in 2005"

Stock market forecasting using hidden Markov model: a new approach

Abstract: This paper presents hidden Markov models (HMM) approach for forecasting stock price for interrelated markets. We apply HMM to forecast some of the airlines stock. HMMs have been extensively used for pattern recognition and classification problems because of its proven suitability for modelling dynamic systems. However, using HMM for predicting future events is not straightforward. Here we use only one HMM that is trained on the past dataset of the chosen airlines. The trained HMM is used to search for the variable of interest behavioural data pattern from the past dataset. By interpolating the neighbouring values of these datasets forecasts are prepared. The results obtained using HMM are encouraging and HMM offers a new paradigm for stock market forecasting, an area that has been of much research interest lately.

A summary-attainment-surface plotting method for visualizing the performance of stochastic multiobjective optimizers

Abstract: When evaluating the performance of a stochastic optimizer it is sometimes desirable to express performance in terms of the quality attained in a certain fraction of sample runs. For example, the sample median quality is the best estimator of what one would expect to achieve in 50% of runs, and similarly for other quantiles. In multiobjective optimization, the notion still applies but the outcome of a run is measured not as a scalar (i.e. the cost of the best solution), but as an attainment surface in k-dimensional space (where k is the number of objectives). In this paper we report an algorithm that can be conveniently used to plot summary attainment surfaces in any number of dimensions (though it is particularly suited for three). A summary attainment surface is defined as the union of all tightest goals that have been attained (independently) in precisely s of the runs of a sample of n runs, for any s/spl isin/1..n, and for any k. We also discuss the computational complexity of the algorithm and give some examples of its use. C code for the algorithm is available from the author.

Applying SVD on item-based filtering

Abstract: In this paper we examine the use of a matrix factorization technique called singular value decomposition (SVD) in item-based collaborative filtering. After a brief introduction to SVD and some of its previous applications in recommender systems, we proceed with a full description of our algorithm, which uses SVD in order to reduce the dimension of the active item's neighborhood. The experimental part of this work first locates the ideal parameter settings for the algorithm, and concludes by contrasting it with plain item-based filtering which utilizes the original, high dimensional neighborhood. The results show that a reduction in the dimension of the item neighborhood is promising, since it does not only tackle some of the recorded problems of recommender systems, but also assists in increasing the accuracy of systems employing it.

Clustering algorithms for bank customer segmentation

Abstract: Market segmentation is one of the most important area of knowledge-based marketing. In banks, it is really a challenging task as data bases are large and multidimensional. In the paper we consider cluster analysis, which is the methodology, the most often applied in this area. We compare clustering algorithms in cases of high dimensionality with noise. We discuss using three algorithms: density based DBSCAN, k-means and based on it two-phase clustering process. We compare algorithms concerning their effectiveness and scalability. Some experiments with exemplary bank data sets are presented.

IDEAS: intrusion detection based on emotional ants for sensors

Abstract: Due to the wide deployment of sensor networks recently security in sensor networks has become a hot research topic. Popular ways to secure a sensor network are by including cryptographic techniques or by safeguarding sensitive information from unauthorized access/manipulation and by implementing efficient intrusion detection mechanisms. This paper proposes a novel ant colony based intrusion detection mechanism which could also keep track of the intruder trials. The IDEAS technique could work in conjunction with the conventional machine learning based intrusion detection techniques to secure the sensor networks. The algorithm is presented and illustrated by simulating a sensor network.

Cellular automata model of pedestrian dynamics for normal and evacuation conditions

Abstract: The knowledge of the issues connected with pedestrian can be very helpful in the proper desing of buildings or other facilities. A simulation should take into consideration both, normal as well as evacuation conditions. Over the recent years several models of pedestrian movement have been created. The first part of the article contains a brief review of some available models. Then, the author propose a model of pedestrians group dynamics in normal and evacuation situations. A 2D cellular automata, multi-agent model is created. The model proposed takes into account some strategic abilities of each agent/pedestrian. In the model, all decisions are made by an agent in correlation with other agents.

Swarm-based sequencing recommendations in e-learning

Abstract: Open and distance learning (ODL) gives learners freedom of time, place and pace of study, putting learner self-direction centre-stage. However, increased responsibility should not come at the price of over-burdening or abandonment of learners as they progress along their learning journey. This paper introduces an approach to recommending the sequencing of e-learning modules for distance learners based on self-organization theory. It describes an architecture which supports the recording, processing and presentation of collective learner behavior designed to create a feedback loop informing learners of successful paths towards the attainment of learning goals. The article includes initial results from a large-scale experiment designed to validate the approach.

Interactive particle swarm optimization

Abstract: It is often desirable to simultaneously handle several objectives and constraints in practical optimization problems. In some cases, these objectives and constraints are non-commensurable and they are not explicitly/mathematically available. For this kind of problems, interactive optimization may be a good approach. Interactive optimization means that a human user evaluates the potential solutions in qualitative way. In recent years evolutionary computation (EC) was applied for interactive optimization, which approach has became known as interactive evolutionary computation (IEC). The aim of this paper is to propose a new interactive optimization method based on particle swarm optimization (PSO). PSO is a relatively new population based optimization approach, whose concept originates from the simulation of simplified social systems. The paper shows that interactive PSO cannot be based on the same concept as IEC because the information sharing mechanism of PSO significantly differs from EC. So this paper proposes an approach which considers the unique attributes of PSO. The proposed algorithm has been implemented in MATLAB (IPSO toolbox) and applied to a case-study of temperature profile design of a batch beer fermenter. The results show that IPSO is an efficient and comfortable interactive optimization algorithm. The developed IPSO toolbox (for Mat-lab) can be downloaded from the Web site of the authors: http://www.fmt.vein.hu/softcomp/ipso.

A hybrid movie recommender system based on neural networks

Abstract: Recently, there has been a lot of speculation among the members of the artificial intelligence community concerning the way AI can help with the problem of successful information search in the reservoirs of knowledge of Internet. Recommender systems provide a solution to this problem by giving individualized recommendations. Content-based and collaborative filtering are usually applied to predict these recommendations. A combination of the results of these two techniques is used in this work in order to construct a system that provides more precise recommendations concerning movies. The MovieLens data set was used to test the proposed hybrid system.

Hybrid fuzzy-genetic algorithm approach for crew grouping

Abstract: Crew grouping is an important problem and formulating a good solution always involves many challenges. For example, grouping soldiers intelligently to tank combat units, we should take into consideration the combined technical proficiency of the soldiers, the amount of military training, the units from which the soldiers come, their service age, personal background, etc. In this paper, we propose a hybrid fuzzy-genetic algorithm (FGA) approach to solve the crew grouping problem. Fuzzy logic based controllers are applied to fine-tune dynamically the crossover and mutation probability in the genetic algorithms, in an attempt to improve the algorithm performance. The FGA approach is compared with the standard genetic algorithm (SGA). Empirical results clearly demonstrates that while the SGA approach gives satisfactory solutions for the problem, the FGA method usually performs significantly better.

A novel fuzzy approach to recognition of online Persian handwriting

Abstract: Fuzzy logic has proved to be a powerful tool to represent imprecise and irregular patterns. This paper presents a novel fuzzy approach for recognizing online Persian (Farsi) handwriting which is also useful for multi-writer environments. In this approach, the representation of handwriting parameters is accomplished by fuzzy linguistic modeling. The representative features are selected to describe the shape of tokens. Fuzzy linguistic terms provide robustness against handwriting variations. The purposed method was run on a database of Persian isolated handwritten characters and achieved a relatively high recognition rate.

Active hypercontours and contextual classification

Abstract: In the paper a concept of active hypercontours as well as a formalized description of contextual classification and relationship between them are presented. The idea of active hypercontours generalizes traditional active contour methods which are extensively developed in image analysis. The proposed concepts can enable an incorporation of techniques specific for traditional contextual and non-contextual classification problems to active contour approach and vice versa.

Adapting particle swarm optimization to stock markets

Abstract: The paper is focused on the development of intelligent decision-making model which is based on the application of artificial neural networks (ANN) and swarm intelligence algorithm. The proposed model generates one-step forward investment decisions. The ANN are used to make the analysis of historical stock returns and to calculate one day forward possible profit, which could be get while following the model proposed decisions concerning the purchase of the stocks. Subsequently the particle swarm optimization (PSO) algorithm is applied for training of ANN. The training of ANN is made through the adjustment of all ANN towards the weights of "global best" ANN. The experimental investigations were made considering different forms of decision-making model: different structure ANN, input variables etc. The paper introduces experimental investigation for the evaluation of decision-making model. The experimental results show that the application of the proposed decision-making model lets to achieve better results than the average of the market.

Representing users in a travel support system

Abstract: We consider the construction and management of user profiles for an agent-based travel support system, with the goal of providing personalized content for individual users of the system. Profiles consist of statements about the "world of travel." Conditional probabilities for each statement model the strengths of user preferences. These probabilities are derived from implicit and explicit observations of user behavior.

Multiobjective optimization using adaptive Pareto archived evolution strategy

Abstract: This paper proposes a novel adaptive representation for evolutionary multiobjective optimization for solving a stock modeling problem. The standard Pareto achieved evolution strategy (PAES) uses real or binary representation for encoding solutions. Adaptive Pareto archived evolution strategy (APAES) uses dynamic alphabets for encoding solutions. APAES is applied for modeling two popular stock indices involving 4 objective functions. Further, two bench mark test functions for multiobjective optimization are also used to illustrate the performance of the algorithm. Empirical results demonstrate APAES performs well when compared to the standard PAES,.

Sharing user models for adaptive hypermedia applications

Abstract: In this paper we present an approach to sharing a user model among several adaptive hypermedia applications. Current adaptive hypermedia applications often realize the user model as the internal part of the system without any possibility to share this model with other applications. We introduce a Web service called UMoWS (user model Web service). The Web service acts as a store of user characteristics represented in pre-defined ontologies that define semantics of the stored knowledge. Access and corresponding privileges to the Web service is managed by the user through the management interface. The user is also allowed to inspect/modify the state of the user model.

The influence of indirect association rules on recommendation ranking lists

Abstract: Association rules are often utilized in Web recommendation systems for creation of suggested items lists. However, lists obtained in this way may be too short. Indirect association rules are introduced to extend classical, direct association rules and supplement the knowledge they contribute. The relation between ranking lists created on the basis of association rules and hyperlinks existing on Web pages has also been examined in the paper.

Fuzzy flight control system for helicopter intelligence in hover

Abstract: This paper relates to a fuzzy flight control system in spot hovering for a single-rotor helicopter PZL Kania. The model of the fuzzy control system was developed on the basis of computer simulation experiments done by the expert's analysis (pilot's knowledge). The helicopter's mathematical model and its fuzzy flight control system were simulated using Matlab software package. In a series of numerous computer simulations the operation of the fuzzy control system was investigated and the system itself was tuned up. Simulation tests have been performed for the helicopter both in fixed hovering, and in hovering with disturbances. Disturbances were related to a gust of wind or to an accidental motion of one of the rudders. Obtained results give the good promise for building the Web simulator.

Smart archive: a component-based data mining application framework

Abstract: Implementation of data mining applications is a challenging and complicated task, and the applications are often built from scratch. In this paper, a component-based application framework, called smart archive (SA) designed for implementing data mining applications, is presented. SA provides functionality common to most data mining applications and components for utilizing history information. Using SA, it is possible to build high-quality applications with shorter development times by configuring the framework to process application-specific data. The architecture, the components, the implementation and the design principles of the framework are presented. The advantages of a framework-based implementation are demonstrated by presenting a case study which compares the framework approach to implementing a real-world application with the option of building an equivalent application from scratch. In conclusion, the paper presents a lucid framework for creating data mining applications and illustrates the importance and advantages of using the presented approach.

A half-eye wavelet based method for iris recognition

Abstract: Iris detection is a crucial part of an iris recognition system. One of the main issues in iris segmentation is coping with occlusions that happen due to eyelids and eyelashes. In this paper, only the lower part of the iris is utilized for recognition. Wavelet based texture features along with a mixed Hamming; harmonic mean distance classifier is used for identification. It is observed that relying in a smaller but more reliable part of the iris, though reducing the net amount of information, improves the overall performance. Experimental results on CASIA database show that the method has a promising performance with an accuracy of more than 99%.

Adaptive fuzzy logic controller with rule-based changeable universe of discourse for a nonlinear MIMO system

Abstract: The accurate input-output universe of discourse (UOD) on which membership functions are defined is hard to acquire, especially for nonlinear multi-input and multi-output (MIMO) systems, and control accuracy will reduce greatly in the steady state due to limited fuzzy control rules. This paper presents a design scheme for the adaptive fuzzy logic controllers to control a MIMO system with online changeable input-output UOD of the controller, which could solve two problems mentioned above to a great extent. We obtain fuzzy inference rules that are defined by the error and change of error to describe the variation law of the input-output UOD according to the current trend of the controlled process, so the proper widths of the UOD can be derived from the fuzzy inference rules dynamically in real-time. Simulation results for the nonlinear MIMO system show the effectiveness of the proposed fuzzy logic controller.

New proposition for intelligent systems design: artificial understanding of the images as the next step of advanced data analysis after automatic classification and pattern recognition

Abstract: In this paper there will be presented the new opportunities for applying linguistic algorithms of pattern recognition for computer understanding of image semantic content in intelligent information systems. A successful obtaining of the crucial semantic information of the image - especially medical - may contribute considerably to the creation of new intelligent cognitive information systems. Thanks to the new algorithms of cognitive resonance between stream of the data extracted from the image and expectations taken from the representation of the medical knowledge, we can understand the merit content of the image even if the form of the image is very different from any known pattern. It seems that in the near future the technique of automatic understanding of images may become one of the effective tools for semantic interpreting, and intelligent storing of the visual data in scattered databases. In this article we will try proving that structural techniques may be applied in the case of tasks related to automatic classification and machine perception of the semantic meaning of selected classes of medical patterns.

An intelligent knowledge sharing strategy featuring item-based collaborative filtering and case based reasoning

Abstract: In this paper, we propose a new approach for combining item-based collaborative filtering (CF) with case based reasoning (CBR) to pursue personalized information filtering in a knowledge sharing context. Functionally, our personalized information filtering approach allows the use of recommendations by peers with similar interests and domain experts to guide the selection of information deemed relevant to an active user's profile. We apply item-based similarity computation in a CF framework to retrieve N information objects based on the user's interests and recommended by peer. The N information objects are then subjected to a CBR based compositional adaptation method to further select relevant information objects from the N retrieved past cases in order to generate a more fine-grained recommendation.

Modern control approach for robotic wheelchair with inverse pendulum control

Abstract: We are proposing a robotic wheelchair that enables a wheelchair bound person to climb over steps up to about 8 cm in height without assistance from others. By using the proposed robotic wheelchair, a user can maintain inverse pendulum control after raising its front wheels. Then, a user can move forward to the step maintaining the inverse pendulum control, and can climb over the step using motor force of a rear wheel shaft. This paper described the control system designs and simulations of the inverse pendulum control. An observer based optimal control (LQG or H/sub 2/) with an integral action is discussed in order to obtain better control performances for the inverse pendulum control.

Finding reducts without building the discernibility matrix

Abstract: We present algorithms for fast generation of short reducts which avoid building the discernibility matrix explicitly. We show how information obtained from this matrix can be obtained based only on the distributions of attribute values. Since the size of discernibility matrix is quadratic in the number of data records, not building the matrix explicitly gives a very significant speedup and makes it possible to find reducts even in very large databases. Algorithms are given for both absolute and relative reducts. Experiments show that our approach outperforms other reduct finding algorithms. Furthermore it is shown that many heuristic reduct finding algorithms using the discernibility matrix in fact select attributes based on their Gini index. A new definition of conditional Gini index is presented, motivated by reduct finding heuristics.

Changeability of Web objects - browser perspective

Abstract: The paper examines the usefulness of the browser cache for studying the properties of Web objects. The cache contains detailed, personalized data on a user interaction with the Web during the period covering a few previous weeks. The data could be used for a variety of purposes including adaptive systems but the paper concentrates on the objects' changeability. Precise knowledge about the scope and nature of the changeability enables us to estimate the upper limit of caching efficiency, no matter what algorithms are used. Caching reduces both the volume of the traffic and the perceived latency. The paper discusses the results of an experiment and suggests other areas in which the data extracted from the local Internet buffer may be useful.

A city guide agent creating and adapting individual sightseeing tours

Abstract: Tourists deciding to explore a destination spontaneously and unprepared will have to walk and search on their own. This kind of investigation can be very uncomfortable as it often ends up in disarrangement. With today's agent technology, tourists can have their own intelligent guide taking care of the whole tour organization and execution in time. This is the main objective of the dynamic tour guide (DTG) - a mobile agent that selects attractions, plans an individual tour, provides navigational guidance and offers location based interpretation. Over all it consistently adapts the tour to a tourist's specific behavior in order to provide any possible support via a mobile device.

Detecting LSB steganography based on dynamic masks

Abstract: This paper presents a dynamic regular groups steganalysis (DRS) algorithm to detect LSB steganography. This algorithm dynamically selects an appropriate mask for each image to reduce the initial bias, and estimates' the LSB embedding message ratio by constructing equations with the statistics of regular groups in image. Experimental results show that this algorithm is more accurate and has a lower missing rate and false1 alarm rate than the conventional RS method and some other powerful steganalysis approaches present recently.

Hybrid intelligent systems for predictive toxicology - a distributed approach

Abstract: The main objective of this paper is to propose a homogeneous approach to represent and process in silico models for predictive toxicology and also to improve the computational representation of developed models by harmonizing new trends in predictive data mining. We propose to combine local and global models as ensemble experts by mixing technologies in hybrid systems in order to improve the prediction accuracy, and also to provide reasonable training response time by using parallel processing. More investigations have still to be done to develop an optimized strategy, but our approach demonstrates encouraging results.

How to use crowding selection in grammar-based classifier system

Abstract: The grammar-based classifier system (GCS) is a new version of learning classifier systems (LCS) in which classifiers are represented by context-free grammar in Chomsky normal form. GCS evolves one grammar during induction (the Michigan approach) which gives it the ability to find the proper set of rules very quickly. However it is quite sensitive to any variations of learning parameters. This paper investigates the role of crowding selection in GCS. To evaluate the performance of GCS depending on crowding factor and crowding subpopulation we used context-free language in the form of so-called toy language. The set of experiments was performed to obtain the answer for question in the title.