Preferred Language
Articles
/
GheZh5IBVTCNdQwCRLMn
Improving Pre-trained CNN-LSTM Models for Image Captioning with Hyper-Parameter Optimization
...Show More Authors

The issue of image captioning, which comprises automatic text generation to understand an image’s visual information, has become feasible with the developments in object recognition and image classification. Deep learning has received much interest from the scientific community and can be very useful in real-world applications. The proposed image captioning approach involves the use of Convolution Neural Network (CNN) pre-trained models combined with Long Short Term Memory (LSTM) to generate image captions. The process includes two stages. The first stage entails training the CNN-LSTM models using baseline hyper-parameters and the second stage encompasses training CNN-LSTM models by optimizing and adjusting the hyper-parameters of the previous stage. Improvements include the use of a new activation function, regular parameter tuning, and an improved learning rate in the later stages of training. The experimental results on the flickr8k dataset showed a noticeable and satisfactory improvement in the second stage, where a clear increment was achieved in the evaluation metrics Bleu1-4, Meteor, and Rouge-L. This increment confirmed the effectiveness of the alterations and highlighted the importance of hyper-parameter tuning in improving the performance of CNN-LSTM models in image caption tasks.

Scopus Crossref
View Publication
Publication Date
Thu Aug 25 2016
Journal Name
International Journal Of Mathematics Trends And Technology
Pretest Single Stage Shrinkage Estimator for the Shape Parameter of the Power Function Distribution
...Show More Authors

View Publication
Crossref (1)
Crossref
Publication Date
Sat Jan 01 2011
Journal Name
Journal Of Engineering
TWO-PARAMETER GAMMA DISTRIBUTION AND LOG NORMAL DISTRIBUTION FOR DERIVATION OF SYNTHETIC UNIT HYDROGRAPH
...Show More Authors

Most available methods for unit hydrographs (SUH) derivation involve manual, subjective fitting of
a hydrograph through a few data points. The use of probability distributions for the derivation of synthetic
hydrographs had received much attention because of its similarity with unit hydrograph properties. In this
paper, the use of two flexible probability distributions is presented. For each distribution the unknown
parameters were derived in terms of the time to peak(tp), and the peak discharge(Qp). A simple Matlab
program is prepared for calculating these parameters and their validity was checked using comparison
with field data. Application to field data shows that the gamma and lognormal distributions had fit well.<

... Show More
View Publication Preview PDF
Crossref (1)
Crossref
Publication Date
Thu Jun 01 2017
Journal Name
Journal Of Economics And Administrative Sciences
Proposed Entropy Loss function and application to find Bayesian estimator for Exponential distribution parameter
...Show More Authors

The aim of this paper to find Bayes estimator under new loss function assemble between symmetric and asymmetric loss functions, namely, proposed entropy loss function, where this function that merge between entropy loss function and the squared Log error Loss function, which is quite asymmetric in nature. then comparison a the Bayes estimators of exponential distribution under the proposed function, whoever, loss functions ingredient for the proposed function the using a standard mean square error (MSE) and Bias quantity (Mbias), where the generation of the random data using the simulation for estimate exponential distribution parameters different sample sizes (n=10,50,100) and (N=1000), taking initial

... Show More
View Publication Preview PDF
Crossref
Publication Date
Sat Oct 01 2016
Journal Name
Journal Of Economics And Administrative Sciences
Bayesian Estimator for the Scale Parameter of the Normal Distribution Under Different Prior Distributions
...Show More Authors

In this study, we used Bayesian method to estimate scale parameter for the normal distribution. By considering three different prior distributions such as the square root inverted gamma (SRIG) distribution and the non-informative prior distribution and the natural conjugate family of priors. The Bayesian estimation based on squared error loss function, and compared it with the classical estimation methods to estimate the scale parameter for the normal distribution, such as the maximum likelihood estimation and th

... Show More
View Publication Preview PDF
Crossref
Publication Date
Sun Jan 01 2012
Journal Name
International Journal Of Contemporary Mathematical Sciences
Pre-Topology Generated by the Short Path Problems
...Show More Authors

Let G be a graph, each edge e of which is given a weight w(e). The shortest path problem is a path of minimum weight connecting two specified vertices a and b, and from it we have a pre-topology. Furthermore, we study the restriction and separators in pre-topology generated by the shortest path problems. Finally, we study the rate of liaison in pre-topology between two subgraphs. It is formally shown that the new distance measure is a metric

Preview PDF
Publication Date
Fri Mar 01 2013
Journal Name
Journal Of Economics And Administrative Sciences
Improving the Effectiveness of Internal Control System In Accordance with COSO Model
...Show More Authors

        The research aims to improve the effectiveness of internal control system according to a model COSO, by identifying the availability of system components according to the model and then improve the effectiveness of each component by focusing on areas for improvement in each component, as it was addressed to a model COSO and then Maamth with the environment, the current Iraqi by introducing some improvements on the form of some mechanisms of corporate governance of the Council of Directors, and senior management, the Audit Committee, Committee appointments, especially that supplies application available in the laws and legislation, the current Iraqi, taking into consideration to make some

... Show More
View Publication Preview PDF
Crossref
Publication Date
Tue Aug 01 2023
Journal Name
International Journal Of Online And Biomedical Engineering (ijoe)
End-to-End Speaker Profiling Using 1D CNN Architectures and Filter Bank Initialization
...Show More Authors

The automatic estimation of speaker characteristics, such as height, age, and gender, has various applications in forensics, surveillance, customer service, and many human-robot interaction applications. These applications are often required to produce a response promptly. This work proposes a novel approach to speaker profiling by combining filter bank initializations, such as continuous wavelets and gammatone filter banks, with one-dimensional (1D) convolutional neural networks (CNN) and residual blocks. The proposed end-to-end model goes from the raw waveform to an estimated height, age, and gender of the speaker by learning speaker representation directly from the audio signal without relying on handcrafted and pre-computed acou

... Show More
View Publication
Scopus (2)
Crossref (1)
Scopus Clarivate Crossref
Publication Date
Sat Nov 12 2016
Journal Name
International Journal Of Mechanical Engineering And Technology (ijmet)
PERFORMANCE OF TWO-WAY NESTING TECHNIQUES FOR SHALLOW WATER MODELS
...Show More Authors

A new two-way nesting technique is presented for a multiple nested-grid ocean modelling system. The new technique uses explicit center finite difference and leapfrog schemes to exchange information between the different subcomponents of the nested-grid system. The performance of the different nesting techniques is compared, using two independent nested-grid modelling systems. In this paper, a new nesting algorithm is described and some preliminary results are demonstrated. The validity of the nesting method is shown in some problems for the depth averaged of 2D linear shallow water equation.

Publication Date
Sun Sep 01 2019
Journal Name
Baghdad Science Journal
Symmetric and Positive Definite Broyden Update for Unconstrained Optimization
...Show More Authors

Broyden update is one of the one-rank updates which solves the unconstrained optimization problem but this update does not guarantee the positive definite and the symmetric property of Hessian matrix.

In this paper the guarantee of positive definite and symmetric property for the Hessian matrix will be established by updating the vector  which represents the difference between the next gradient and the current gradient of the objective function assumed to be twice continuous and differentiable .Numerical results are reported to compare the proposed method with the Broyden method under standard problems.

View Publication Preview PDF
Scopus (10)
Crossref (2)
Scopus Clarivate Crossref
Publication Date
Thu Nov 02 2023
Journal Name
Journal Of Engineering
An Improved Adaptive Spiral Dynamic Algorithm for Global Optimization
...Show More Authors

This paper proposes a new strategy to enhance the performance and accuracy of the Spiral dynamic algorithm (SDA) for use in solving real-world problems by hybridizing the SDA with the Bacterial Foraging optimization algorithm (BFA). The dynamic step size of SDA makes it a useful exploitation approach. However, it has limited exploration throughout the diversification phase, which results in getting trapped at local optima. The optimal initialization position for the SDA algorithm has been determined with the help of the chemotactic strategy of the BFA optimization algorithm, which has been utilized to improve the exploration approach of the SDA. The proposed Hybrid Adaptive Spiral Dynamic Bacterial Foraging (HASDBF)

... Show More
View Publication Preview PDF
Crossref