Abstract: Water level forecasting using records of past time series is of importance in water resources engineering and management. For example, water level affects groundwater tables in low-lying coastal areas, as well as hydrological regimes of some coastal rivers. Then, a reliable prediction of sea-level variations is required in coastal engineering and hydrologic studies. During the past two decades, the approaches based on the Genetic Programming (GP) and Artificial Neural Networks (ANN) were developed. In the present study, the GP is used to forecast daily water level variations for a set of time intervals using observed water levels. The measurements from a single tide gauge at Urmia Lake, Northwest Iran, were used to train and validate the GP approach for the period from January 1997 to July 2008. Statistics, the root mean square error and correlation coefficient, are used to verify model by comparing with a corresponding outputs from Artificial Neural Network model. The results show that both these artificial intelligence methodologies are satisfactory and can be considered as alternatives to the conventional harmonic analysis.
Abstract: This paper presents the novel Rao-Blackwellised
particle filter (RBPF) for mobile robot simultaneous localization and
mapping (SLAM) using monocular vision. The particle filter is
combined with unscented Kalman filter (UKF) to extending the path
posterior by sampling new poses that integrate the current observation
which drastically reduces the uncertainty about the robot pose. The
landmark position estimation and update is also implemented through
UKF. Furthermore, the number of resampling steps is determined
adaptively, which seriously reduces the particle depletion problem,
and introducing the evolution strategies (ES) for avoiding particle
impoverishment. The 3D natural point landmarks are structured with
matching Scale Invariant Feature Transform (SIFT) feature pairs. The
matching for multi-dimension SIFT features is implemented with a
KD-Tree in the time cost of O(log2
N). Experiment results on real robot
in our indoor environment show the advantages of our methods over
previous approaches.
Abstract: In this paper we present a GP-based method for automatically evolve projections, so that data can be more easily classified in the projected spaces. At the same time, our approach can reduce dimensionality by constructing more relevant attributes. Fitness of each projection measures how easy is to classify the dataset after applying the projection. This is quickly computed by a Simple Linear Perceptron. We have tested our approach in three domains. The experiments show that it obtains good results, compared to other Machine Learning approaches, while reducing dimensionality in many cases.
Abstract: Iris-based biometric system is gaining its importance in several applications. However, processing of iris biometric is a challenging and time consuming task. Detection of iris part in an eye image poses a number of challenges such as, inferior image quality, occlusion of eyelids and eyelashes etc. Due to these problems it is not possible to achieve 100% accuracy rate in any iris-based biometric authentication systems. Further, iris detection is a computationally intensive task in the overall iris biometric processing. In this paper, we address these two problems and propose a technique to localize iris part efficiently and accurately. We propose scaling and color level transform followed by thresholding, finding pupil boundary points for pupil boundary detection and dilation, thresholding, vertical edge detection and removal of unnecessary edges present in the eye images for iris boundary detection. Scaling reduces the search space significantly and intensity level transform is helpful for image thresholding. Experimental results show that our approach is comparable with the existing approaches. Following our approach it is possible to detect iris part with 95-99% accuracy as substantiated by our experiments on CASIA Ver-3.0, ICE 2005, UBIRIS, Bath and MMU iris image databases.
Abstract: As the number of networked computers grows,
intrusion detection is an essential component in keeping networks
secure. Various approaches for intrusion detection are currently
being in use with each one has its own merits and demerits. This
paper presents our work to test and improve the performance of a
new class of decision tree c-fuzzy decision tree to detect intrusion.
The work also includes identifying best candidate feature sub set to
build the efficient c-fuzzy decision tree based Intrusion Detection
System (IDS). We investigated the usefulness of c-fuzzy decision
tree for developing IDS with a data partition based on horizontal
fragmentation. Empirical results indicate the usefulness of our
approach in developing the efficient IDS.
Abstract: During the past several years, face recognition in video
has received significant attention. Not only the wide range of
commercial and law enforcement applications, but also the availability
of feasible technologies after several decades of research contributes
to the trend. Although current face recognition systems have reached a
certain level of maturity, their development is still limited by the
conditions brought about by many real applications. For example,
recognition images of video sequence acquired in an open
environment with changes in illumination and/or pose and/or facial
occlusion and/or low resolution of acquired image remains a largely
unsolved problem. In other words, current algorithms are yet to be
developed. This paper provides an up-to-date survey of video-based
face recognition research. To present a comprehensive survey, we
categorize existing video based recognition approaches and present
detailed descriptions of representative methods within each category.
In addition, relevant topics such as real time detection, real time
tracking for video, issues such as illumination, pose, 3D and low
resolution are covered.
Abstract: Sign language recognition has been a topic of research since the first data glove was developed. Many researchers have attempted to recognize sign language through various techniques. However none of them have ventured into the area of Pakistan Sign Language (PSL). The Boltay Haath project aims at recognizing PSL gestures using Statistical Template Matching. The primary input device is the DataGlove5 developed by 5DT. Alternative approaches use camera-based recognition which, being sensitive to environmental changes are not always a good choice.This paper explains the use of Statistical Template Matching for gesture recognition in Boltay Haath. The system recognizes one handed alphabet signs from PSL.
Abstract: In order to enhance the contrast in the regions where the pixels have similar intensities, this paper presents a new histogram equalization scheme. Conventional global equalization schemes over-equalizes these regions so that too bright or dark pixels are resulted and local equalization schemes produce unexpected discontinuities at the boundaries of the blocks. The proposed algorithm segments the original histogram into sub-histograms with reference to brightness level and equalizes each sub-histogram with the limited extents of equalization considering its mean and variance. The final image is determined as the weighted sum of the equalized images obtained by using the sub-histogram equalizations. By limiting the maximum and minimum ranges of equalization operations on individual sub-histograms, the over-equalization effect is eliminated. Also the result image does not miss feature information in low density histogram region since the remaining these area is applied separating equalization. This paper includes how to determine the segmentation points in the histogram. The proposed algorithm has been tested with more than 100 images having various contrasts in the images and the results are compared to the conventional approaches to show its superiority.
Abstract: Automatic reusability appraisal is helpful in
evaluating the quality of developed or developing reusable software
components and in identification of reusable components from
existing legacy systems; that can save cost of developing the
software from scratch. But the issue of how to identify reusable
components from existing systems has remained relatively
unexplored. In this research work, structural attributes of software
components are explored using software metrics and quality of the
software is inferred by different Neural Network based approaches,
taking the metric values as input. The calculated reusability value
enables to identify a good quality code automatically. It is found that
the reusability value determined is close to the manual analysis used
to be performed by the programmers or repository managers. So, the
developed system can be used to enhance the productivity and
quality of software development.
Abstract: In this study, a fuzzy similarity approach for Arabic
web pages classification is presented. The approach uses a fuzzy
term-category relation by manipulating membership degree for the
training data and the degree value for a test web page. Six measures
are used and compared in this study. These measures include:
Einstein, Algebraic, Hamacher, MinMax, Special case fuzzy and
Bounded Difference approaches. These measures are applied and
compared using 50 different Arabic web pages. Einstein measure was
gave best performance among the other measures. An analysis of
these measures and concluding remarks are drawn in this study.
Abstract: This paper compares six approaches of object serialization
from qualitative and quantitative aspects. Those are object
serialization in Java, IDL, XStream, Protocol Buffers, Apache Avro,
and MessagePack. Using each approach, a common example is
serialized to a file and the size of the file is measured. The qualitative
comparison works are investigated in the way of checking whether
schema definition is required or not, whether schema compiler is
required or not, whether serialization is based on ascii or binary, and
which programming languages are supported. It is clear that there
is no best solution. Each solution makes good in the context it was
developed.
Abstract: Many research works are carried out on the analysis of
traces in a digital learning environment. These studies produce large
volumes of usage tracks from the various actions performed by a
user. However, to exploit these data, compare and improve
performance, several issues are raised. To remedy this, several works
deal with this problem seen recently. This research studied a series of
questions about format and description of the data to be shared. Our
goal is to share thoughts on these issues by presenting our experience
in the analysis of trace-based log files, comparing several approaches
used in automatic classification applied to e-learning platforms.
Finally, the obtained results are discussed.
Abstract: Text document categorization involves large amount
of data or features. The high dimensionality of features is a
troublesome and can affect the performance of the classification.
Therefore, feature selection is strongly considered as one of the
crucial part in text document categorization. Selecting the best
features to represent documents can reduce the dimensionality of
feature space hence increase the performance. There were many
approaches has been implemented by various researchers to
overcome this problem. This paper proposed a novel hybrid approach
for feature selection in text document categorization based on Ant
Colony Optimization (ACO) and Information Gain (IG). We also
presented state-of-the-art algorithms by several other researchers.
Abstract: In this paper, we propose a Connect6 solver which
adopts a hybrid approach based on a tree-search algorithm and image
processing techniques. The solver must deal with the complicated
computation and provide high performance in order to make real-time
decisions. The proposed approach enables the solver to be
implemented on a single Spartan-6 XC6SLX45 FPGA produced by
XILINX without using any external devices. The compact
implementation is achieved through image processing techniques to
optimize a tree-search algorithm of the Connect6 game. The tree
search is widely used in computer games and the optimal search brings
the best move in every turn of a computer game. Thus, many
tree-search algorithms such as Minimax algorithm and artificial
intelligence approaches have been widely proposed in this field.
However, there is one fundamental problem in this area; the
computation time increases rapidly in response to the growth of the
game tree. It means the larger the game tree is, the bigger the circuit
size is because of their highly parallel computation characteristics.
Here, this paper aims to reduce the size of a Connect6 game tree using
image processing techniques and its position symmetric property. The
proposed solver is composed of four computational modules: a
two-dimensional checkmate strategy checker, a template matching
module, a skilful-line predictor, and a next-move selector. These
modules work well together in selecting next moves from some
candidates and the total amount of their circuits is small. The details of
the hardware design for an FPGA implementation are described and
the performance of this design is also shown in this paper.
Abstract: In recent years multi-agent systems have emerged as one of the interesting architectures facilitating distributed collaboration and distributed problem solving. Each node (agent) of the network might pursue its own agenda, exploit its environment, develop its own problem solving strategy and establish required communication strategies. Within each node of the network, one could encounter a diversity of problem-solving approaches. Quite commonly the agents can realize their processing at the level of information granules that is the most suitable from their local points of view. Information granules can come at various levels of granularity. Each agent could exploit a certain formalism of information granulation engaging a machinery of fuzzy sets, interval analysis, rough sets, just to name a few dominant technologies of granular computing. Having this in mind, arises a fundamental issue of forming effective interaction linkages between the agents so that they fully broadcast their findings and benefit from interacting with others.
Abstract: It has been defined that the “network is the system".
This implies providing levels of service, reliability, predictability and
availability that are commensurate with or better than those that
individual computers provide today. To provide this requires
integrated network management for interconnected networks of
heterogeneous devices covering both the local campus. In this paper
we are addressing a framework to effectively deal with this issue. It
consists of components and interactions between them which are
required to perform the service fault management. A real-world
scenario is used to derive the requirements which have been applied
to the component identification. An analysis of existing frameworks
and approaches with respect to their applicability to the framework is
also carried out.
Abstract: This paper presents a new problem solving approach
that is able to generate optimal policy solution for finite-state
stochastic sequential decision-making problems with high data
efficiency. The proposed algorithm iteratively builds and improves
an approximate Markov Decision Process (MDP) model along with
cost-to-go value approximates by generating finite length trajectories
through the state-space. The approach creates a synergy between an
approximate evolving model and approximate cost-to-go values to
produce a sequence of improving policies finally converging to the
optimal policy through an intelligent and structured search of the
policy space. The approach modifies the policy update step of the
policy iteration so as to result in a speedy and stable convergence to
the optimal policy. We apply the algorithm to a non-holonomic
mobile robot control problem and compare its performance with
other Reinforcement Learning (RL) approaches, e.g., a) Q-learning,
b) Watkins Q(λ), c) SARSA(λ).
Abstract: The article describes problems of city centers with regard to possibilities of their delimitation in a GIS environment. First the definitions and delimitations of a city centre which are in use are mentioned, furthermore a chosen case study (the historical centre of Olomouc city in the Czech Republic) is employed to describe the methods of delimitation in use. In addition to describing the current state, the article also deals with possibilities of delimitation of a city centre in GIS environment by means of several chosen approaches. The authors describe, compare and discuss the chosen methods and assess the achieved results and also applicability of the designed methods for other cities.
Abstract: In this paper a comprehensive model of a fossil fueled
power plant (FFPP) is developed in order to evaluate the
performance of a newly designed turbine follower controller.
Considering the drawbacks of previous works, an overall model is
developed to minimize the error between each subsystem model
output and the experimental data obtained at the actual power plant.
The developed model is organized in two main subsystems namely;
Boiler and Turbine. Considering each FFPP subsystem
characteristics, different modeling approaches are developed. For
economizer, evaporator, superheater and reheater, first order models
are determined based on principles of mass and energy conservation.
Simulations verify the accuracy of the developed models. Due to the
nonlinear characteristics of attemperator, a new model, based on a
genetic-fuzzy systems utilizing Pittsburgh approach is developed
showing a promising performance vis-à-vis those derived with other
methods like ANFIS. The optimization constraints are handled
utilizing penalty functions. The effect of increasing the number of
rules and membership functions on the performance of the proposed
model is also studied and evaluated. The turbine model is developed
based on the equation of adiabatic expansion. Parameters of all
evaluated models are tuned by means of evolutionary algorithms.
Based on the developed model a fuzzy PI controller is developed. It
is then successfully implemented in the turbine follower control
strategy of the plant. In this control strategy instead of keeping
control parameters constant, they are adjusted on-line with regard to
the error and the error rate. It is shown that the response of the
system improves significantly. It is also shown that fuel consumption
decreases considerably.
Abstract: Modeling of complex dynamic systems, which are
very complicated to establish mathematical models, requires new and
modern methodologies that will exploit the existing expert
knowledge, human experience and historical data. Fuzzy cognitive
maps are very suitable, simple, and powerful tools for simulation and
analysis of these kinds of dynamic systems. However, human experts
are subjective and can handle only relatively simple fuzzy cognitive
maps; therefore, there is a need of developing new approaches for an
automated generation of fuzzy cognitive maps using historical data.
In this study, a new learning algorithm, which is called Big Bang-Big
Crunch, is proposed for the first time in literature for an automated
generation of fuzzy cognitive maps from data. Two real-world
examples; namely a process control system and radiation therapy
process, and one synthetic model are used to emphasize the
effectiveness and usefulness of the proposed methodology.