Abstract: One major source of performance decline in speaker
recognition system is channel mismatch between training and testing.
This paper focuses on improving channel robustness of speaker
recognition system in two aspects of channel compensation technique
and channel robust features. The system is text-independent speaker
identification system based on two-stage recognition. In the aspect of
channel compensation technique, this paper applies MAP (Maximum
A Posterior Probability) channel compensation technique, which was
used in speech recognition, to speaker recognition system. In the
aspect of channel robust features, this paper introduces
pitch-dependent features and pitch-dependent speaker model for the
second stage recognition. Based on the first stage recognition to
testing speech using GMM (Gaussian Mixture Model), the system
uses GMM scores to decide if it needs to be recognized again. If it
needs to, the system selects a few speakers from all of the speakers
who participate in the first stage recognition for the second stage
recognition. For each selected speaker, the system obtains 3
pitch-dependent results from his pitch-dependent speaker model, and
then uses ANN (Artificial Neural Network) to unite the 3
pitch-dependent results and 1 GMM score for getting a fused result.
The system makes the second stage recognition based on these fused
results. The experiments show that the correct rate of two-stage
recognition system based on MAP channel compensation technique
and pitch-dependent features is 41.7% better than the baseline system
for closed-set test.
Abstract: This paper tries to shed light on the existence of a bank lending channel (BLC) in South Eastern European countries (SEE). Based on a VAR framework we test the responsiveness of credit supply to monetary policy shocks. By compiling a new data set and using the reserve requirement ratio, among others, as the policy instrument we measure the effectiveness of the BLC and the buffering effect of the banks in the SEE countries. The results indicate that loan supply is significantly affected by shifts in monetary policy, when demand factors are controlled. Furthermore, by analyzing the effect of the Greek banks in the region we conclude that Greek banks do buffer the negative effects of monetary policy transmission. By having a significant market share of the SEE-s banking markets we argue that Greek banks influence positively the economic growth of SEE countries.
Abstract: This paper presents a new classification algorithm using colour and texture for obstacle detection. Colour information is computationally cheap to learn and process. However in many cases, colour alone does not provide enough information for classification. Texture information can improve classification performance but usually comes at an expensive cost. Our algorithm uses both colour and texture features but texture is only needed when colour is unreliable. During the training stage, texture features are learned specifically to improve the performance of a colour classifier. The algorithm learns a set of simple texture features and only the most effective features are used in the classification stage. Therefore our algorithm has a very good classification rate while is still fast enough to run on a limited computer platform. The proposed algorithm was tested with a challenging outdoor image set. Test result shows the algorithm achieves a much better trade-off between classification performance and efficiency than a typical colour classifier.
Abstract: Mobile Ad hoc network consists of a set of mobile
nodes. It is a dynamic network which does not have fixed topology.
This network does not have any infrastructure or central
administration, hence it is called infrastructure-less network. The
change in topology makes the route from source to destination as
dynamic fixed and changes with respect to time. The nature of
network requires the algorithm to perform route discovery, maintain
route and detect failure along the path between two nodes [1]. This
paper presents the enhancements of ARA [2] to improve the
performance of routing algorithm. ARA [2] finds route between
nodes in mobile ad-hoc network. The algorithm is on-demand source
initiated routing algorithm. This is based on the principles of swarm
intelligence. The algorithm is adaptive, scalable and favors load
balancing. The improvements suggested in this paper are handling of
loss ants and resource reservation.
Abstract: This paper mathematically analyses the varying
magnitude of production loss, which may occur due to idle time (inprocess
waiting time and traveling time) on a linear walking worker
assembly line. Within this flexible and reconfigurable assembly
system, each worker travels down the line carrying out each
assembly task at each station; and each worker accomplishes the
assembly of a unit from start to finish and then travels back to the
first station to start the assembly of a new product. This strategy of
system design attempts to combine the flexibility of the U-shaped
moving worker assembly cell with the efficiency of the conventional
fixed worker assembly line. The paper aims to evaluate the effect of
idle time that may offset the labor efficiency of each walking worker
providing an insight into the mechanism of such a flexible and
reconfigurable assembly system.
Abstract: Probabilistic techniques in computer programs are becoming
more and more widely used. Therefore, there is a big
interest in the formal specification, verification, and development
of probabilistic programs. In our work-in-progress project, we are
attempting to make a constructive framework for developing probabilistic
programs formally. The main contribution of this paper
is to introduce an intermediate artifact of our work, a Z-based
formalism called PZ, by which one can build set theoretical models of
probabilistic programs. We propose to use a constructive set theory,
called CZ set theory, to interpret the specifications written in PZ.
Since CZ has an interpretation in Martin-L¨of-s theory of types, this
idea enables us to derive probabilistic programs from correctness
proofs of their PZ specifications.
Abstract: The purpose of this paper is to describe the process of
setting up a learning community within an elementary school in
Ontario, Canada. The description is provided through reflection and
examination of field notes taken during the yearlong training and
implementation process. Specifically the impact of teachers- capacity
on the creation of a learning community was of interest. This paper is
intended to inform and add to the debate around the tensions that
exist in implementing a bottom-up professional development model
like the learning community in a top-down organizational structure.
My reflections of the process illustrate that implementation of the
learning community professional development model may be
difficult and yet transformative in the professional lives of the
teachers, students, and administration involved in the change process.
I conclude by suggesting the need for a new model of professional
development that requires a transformative shift in power dynamics
and a shift in the view of what constitutes effective professional
learning.
Abstract: There are multiple reasons to expect that detecting the
word order errors in a text will be a difficult problem, and detection
rates reported in the literature are in fact low. Although grammatical
rules constructed by computer linguists improve the performance of
grammar checker in word order diagnosis, the repairing task is still
very difficult. This paper presents an approach for repairing word
order errors in English text by reordering words in a sentence and
choosing the version that maximizes the number of trigram hits
according to a language model. The novelty of this method concerns
the use of an efficient confusion matrix technique for reordering the
words. The comparative advantage of this method is that works with
a large set of words, and avoids the laborious and costly process of
collecting word order errors for creating error patterns.
Abstract: In open settings, the participants in virtual
organization are autonomous and there is no central authority to
ensure the felicity of their interactions. When agents interact in such
settings, each relies upon being able to model the trustworthiness of
the agents with whom it interacts. Fundamentally, such models must
consider the past behavior of the other parties in order to predict their
future behavior. Further, it is sensible for the agents to share
information via referrals to trustworthy agents. In this article, trust is
a bet on the future contingent actions of others" and enumerates six
major factors supporting it: (1) reputation, (2) performance, (3)
appearance, (4) accountability, (5) precommitment, and (6)
contextual facilitation.
Abstract: UML is a collection of notations for capturing a software system specification. These notations have a specific syntax defined by the Object Management Group (OMG), but many of their constructs only present informal semantics. They are primarily graphical, with textual annotation. The inadequacies of standard UML as a vehicle for complete specification and implementation of real-time embedded systems has led to a variety of competing and complementary proposals. The Real-time UML profile (UML-RT), developed and standardized by OMG, defines a unified framework to express the time, scheduling and performance aspects of a system. We present in this paper a framework approach aimed at deriving a complete specification of a real-time system. Therefore, we combine two methods, a semiformal one, UML-RT, which allows the visual modeling of a realtime system and a formal one, CSP+T, which is a design language including the specification of real-time requirements. As to show the applicability of the approach, a correct design of a real-time system with hard real time constraints by applying a set of mapping rules is obtained.
Abstract: The amount of the information being churned out by the field of biology has jumped manifold and now requires the extensive use of computer techniques for the management of this information. The predominance of biological information such as protein sequence similarity in the biological information sea is key information for detecting protein evolutionary relationship. Protein sequence similarity typically implies homology, which in turn may imply structural and functional similarities. In this work, we propose, a learning method for detecting remote protein homology. The proposed method uses a transformation that converts protein sequence into fixed-dimensional representative feature vectors. Each feature vector records the sensitivity of a protein sequence to a set of amino acids substrings generated from the protein sequences of interest. These features are then used in conjunction with support vector machines for the detection of the protein remote homology. The proposed method is tested and evaluated on two different benchmark protein datasets and it-s able to deliver improvements over most of the existing homology detection methods.
Abstract: We propose an all optical flip-flop circuit composedof two Silicon-on-insulator microring resonators coupled to straightwaveguides by exploiting the optical bistability behavior due to thenonlinear Kerr effect. We used the transfer matrix analysis toinvestigate continuous wave propagation through microrings, as wellwe considered the nonlinear switching characteristics of an opticaldevice using a double-coupler silicon ring resonator in presence ofthe Kerr nonlinearity, thus obtaining the bistability behavior of theoutput port, the drop port and also inside the silicon microringresonator. It is shown that the bistability behavior depends on thecontrol of the input wavelength.KeywordsAll optical flip-flops, Kerr effect, microringresonator, optical bistability.
Abstract: The objective is to split a simply connected polygon
into a set of convex quadrilaterals without inserting new
boundary nodes. The presented approach consists in repeatedly
removing quadrilaterals from the polygon. Theoretical results
pertaining to quadrangulation of simply connected polygons are
derived from the usual 2-ear theorem. It produces a quadrangulation
technique with O(n) number of quadrilaterals. The
theoretical methodology is supplemented by practical results
and CAD surface segmentation.
Abstract: In this paper, The T-G-action topology on a set acted
on by a fuzzy T-neighborhood (T-neighborhood, for short) group is
defined as a final T-neighborhood topology with respect to a set of
maps. We mainly prove that this topology is a T-regular Tneighborhood
topology.
Abstract: Tolerance is a tool for achieving a social cohesion, particularly, among individuals and groups with different values. The aim is to study the characteristics of the ethnic tolerance, the inhabitants of Latvia. The ethnic tolerance is taught as a set of conscious and unconscious orientations of the individual in social interaction and inter-ethnic communication. It uses the tools of empirical studies of the ethnic tolerance which allows to identify the explicitly and implicitly levels of the emotional component of Latvia's residents. Explicit measurements were made using the techniques of self-report which revealed the index of the ethnic tolerance and the ethnic identity of the participants. The implicit component was studied using methods based on the effect of the emotional priming. During the processing of the results, there were calculated indicators of the positive and negative implicit attitudes towards members of their own and other ethnicity as well as the explicit parameters of the ethnic tolerance and the ethnic identity of Latvia-s residents. The implicit measurements of the ratio of neighboring ethnic groups against each other showed a mutual negative attitude whereas the explicit measurements indicate a neutral attitude. The data obtained contribute to a further study of the ethnic tolerance of Latvia's residents.
Abstract: The recognition of handwritten numeral is an
important area of research for its applications in post office, banks
and other organizations. This paper presents automatic recognition of
handwritten Kannada numerals based on structural features. Five
different types of features, namely, profile based 10-segment string,
water reservoir; vertical and horizontal strokes, end points and
average boundary length from the minimal bounding box are used in
the recognition of numeral. The effect of each feature and their
combination in the numeral classification is analyzed using nearest
neighbor classifiers. It is common to combine multiple categories of
features into a single feature vector for the classification. Instead,
separate classifiers can be used to classify based on each visual
feature individually and the final classification can be obtained based
on the combination of separate base classification results. One
popular approach is to combine the classifier results into a feature
vector and leaving the decision to next level classifier. This method
is extended to extract a better information, possibility distribution,
from the base classifiers in resolving the conflicts among the
classification results. Here, we use fuzzy k Nearest Neighbor (fuzzy
k-NN) as base classifier for individual feature sets, the results of
which together forms the feature vector for the final k Nearest
Neighbor (k-NN) classifier. Testing is done, using different features,
individually and in combination, on a database containing 1600
samples of different numerals and the results are compared with the
results of different existing methods.
Abstract: The aim of this study was to determine noise level of
six different types of machines in printing companies in Novi Sad.
The A-weighted levels on Leq, Lmax and Lmin Sound Pressure Level
(SPL) in dBA were measured. It was found that the folders, offset
printing presses and binding machines are the predominant noise
sources. The noise levels produced by 12 of 38 machines exceed the
limiting threshold level of 85 dBA, tolerated by law. Since it was
determined that the average noise level for folders (87.7 dB) exceeds
the permitted value the octave analysis of noise was performed.
Abstract: Finding synchronizing sequences for the finite automata is a very important problem in many practical applications (part orienters in industry, reset problem in biocomputing theory, network issues etc). Problem of finding the shortest synchronizing sequence is NP-hard, so polynomial algorithms probably can work only as heuristic ones. In this paper we propose two versions of polynomial algorithms which work better than well-known Eppstein-s Greedy and Cycle algorithms.
Abstract: Mixed Model Production is the practice of assembling
several distinct and different models of a product on the same
assembly line without changeovers and then sequencing those models
in a way that smoothes the demand for upstream components. In this
paper, we consider an objective function which minimizes total
stoppage time and total idle time and it is presented sequence
dependent set up time. Many studies have been done on the mixed
model assembly lines. But in this paper we specifically focused on
reducing the idle times. This is possible through various help policies.
For improving the solutions, some cases developed and about 40 tests
problem was considered. We use scatter search for optimization and
for showing the efficiency of our algorithm, experimental results
shows behavior of method. Scatter search and help policies can
produce high quality answers, so it has been used in this paper.
Abstract: The paper presents the design concept of a unitselection
text-to-speech synthesis system for the Slovenian language.
Due to its modular and upgradable architecture, the system can be
used in a variety of speech user interface applications, ranging from
server carrier-grade voice portal applications, desktop user interfaces
to specialized embedded devices.
Since memory and processing power requirements are important
factors for a possible implementation in embedded devices, lexica
and speech corpora need to be reduced. We describe a simple and
efficient implementation of a greedy subset selection algorithm that
extracts a compact subset of high coverage text sentences. The
experiment on a reference text corpus showed that the subset
selection algorithm produced a compact sentence subset with a small
redundancy.
The adequacy of the spoken output was evaluated by several
subjective tests as they are recommended by the International
Telecommunication Union ITU.