Pronunciation lexicons and prediction models are a key component in several speech synthesis and recognition systems. We know that morphologically related words typically follow a fixed pattern of pronunciation which can be described by language-spec
ific paradigms. In this work we explore how deep recurrent neural networks can be used to automatically learn and exploit this pattern to improve the pronunciation prediction quality of words related by morphological inflection. We propose two novel approaches for supplying morphological information, using the word's morphological class and its lemma, which are typically annotated in standard lexicons. We report improvements across a number of European languages with varying degrees of phonological and morphological complexity, and two language families, with greater improvements for languages where the pronunciation prediction task is inherently more challenging. We also observe that combining bidirectional LSTM networks with attention mechanisms is an effective neural approach for the computational problem considered, across languages. Our approach seems particularly beneficial in the low resource setting, both by itself and in conjunction with transfer learning.
The state-of-the-art on basic, single-antecedent anaphora has greatly improved in recent years. Researchers have therefore started to pay more attention to more complex cases of anaphora such as split-antecedent anaphora, as in Time-Warner is conside
ring a legal challenge to Telecommunications Inc's plan to buy half of Showtime Networks Inc--a move that could lead to all-out war between the two powerful companies''. Split-antecedent anaphora is rarer and more complex to resolve than single-antecedent anaphora; as a result, it is not annotated in many datasets designed to test coreference, and previous work on resolving this type of anaphora was carried out in unrealistic conditions that assume gold mentions and/or gold split-antecedent anaphors are available. These systems also focus on split-antecedent anaphors only. In this work, we introduce a system that resolves both single and split-antecedent anaphors, and evaluate it in a more realistic setting that uses predicted mentions. We also start addressing the question of how to evaluate single and split-antecedent anaphors together using standard coreference evaluation metrics.
This study aims is to analyze the effect of spatial accuracy of the control points on the
images geometric correction accuracy, and this is done by applying tests on the same
image (IKONOS), where polynomial transformations were applied using sets
of control
points, each with absolute accuracy different from the other. These points were
extrapolated from a 1/1000 topographic map and from a georeferenced MOMS satellite
image with geometric accuracy of 2m and measured by GPS. The study showed that it is
possible to obtain the most accurate geometric correction by using control points with
absolute accuracy close to the spatial resolution of the image. It also showed that the use of
more precise control points would not ameliorate the accuracy of the geometric correction,
because the measurement of these points on the image is limited by its spatial resolution.
The main objective of this research is to study the effect of the accuracy of images'
geometric resolution only on the geometric quality of the resulted three-dimensional
model. In this research, all factors that affect the quality of the model are
fixed and the
geometric resolution is changed only for the used images.
The number of captured images, the number and the distribution and the accuracy of
control points, the camera being used and whether or not it is calibrated, are among the
most important factors influencing the modeling process. In order to neutralize the effect of
the inner parameters of the used camera, a process of calibration was achieved. On the
other hand, we have pre-planned the process of photography to avoid problems resulting
from the lack or increase the number of images, that directly affect the quality and
completeness of the model. In addition, accurate control data obtained from precise survey
work (horizontal geodetic network and leveling network) was applied.
In this study, we examined the effect of image resolution on the generation of a dense
cloud of points by applying the Structure from Motion (SfM) and deducing the surface
model and the orthophoto of a facade of a building at Tishreen University.
In this study, we focused on the interpolation methods for the
derivation of digital elevation models, based on field gridding
observations with different spacing.
The present study aimed to discuss the concept of quality of
information provided by the MIS in the Syrian Telecommunications
company with identifying the most important dimensions of
information quality, and how to measure them in order to test their
impact on the decision making process.
This research aims to show the importance of ensuring the
competence of all who operate specific equipment, perform tests
and/or calibrations, evaluate results, and sign test reports and
calibration certificates.
The programming interface presented in this research makes it possible
to manipulate them easily and flexibly by all specialist users, as well as
the possibility of managing, displaying, searching, amending and saving
these plans through the appro
ved programming interface which was
applied to more than one real estate area. It is considered the foundation
in the future comprehensive automation process of the Real Estate
Department. In this research, the same real estate plotting approved by
the General Real Estate Department was adopted ( real estate areas
names and numbers) when manipulating them, using the ARC GIS 9
program.
The purpose of this study, is to explain the factors that may be able to affect the
prices of stocks in the IPO. So it could be priced at it's fair price, not less or more than it.
Which caused by the existence of asymmetric information between the
investors in the
financial market. Whereas, the asymmetric information generates adverse selection with
the investors, about the real values of the company and it's stocks in the IPO market.
This research aims to study the multiple signals, which could be used by the firm to
reduce the asymmetric information and highlight the real value of the firm and it's quality,
in Damascus Securities Exchange between (2009-2011). Some of these signals, which has
been considered in this research are: the standards of Financial and Non Financial
disclosure in it's prospectus.
The researcher reached to several results:
The Financial disclosure standards has no relation to the stock price in the Syrian
IPO market, and cannot be used as a signal to show the real value of the firm. On the other
hand Non-Financial disclosure standards could be used as a good signal to show the real
and fair price of the firms and their stocks in the IPO process.
This result could help the financial market and the firms who want to go public to
use this factors as a signal about it's quality, and for reduce the under pricing in their IPOs.
Studying of land use changing Detection needs to the speed of implementation to
convoy the changes on the ground. the traditional ways in the analysis and visual
interpretation of the images and field studying need a lot of time and effort. So The
objective of this search is to classify group images (Landsat TM , ETM+) taken in different
dates automatically, and then to calculate the area of each land use/land cover, during the
years studied (1990 -2000- 2010) and comparison areas to identify the most important
changes occurring during that period.