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
Mon Jan 01 2024
Journal Name
Lecture Notes On Data Engineering And Communications Technologies
Utilizing Deep Learning Technique for Arabic Image Captioning
...Show More Authors

View Publication
Scopus Crossref
Publication Date
Fri Mar 29 2024
Journal Name
Iraqi Journal Of Science
Evaluating the Performance and Behavior of CNN, LSTM, and GRU for Classification and Prediction Tasks
...Show More Authors

     Deep learning (DL) plays a significant role in several tasks, especially classification and prediction. Classification tasks can be efficiently achieved via convolutional neural networks (CNN) with a huge dataset, while recurrent neural networks (RNN) can perform prediction tasks due to their ability to remember time series data. In this paper, three models have been proposed to certify the evaluation track for classification and prediction tasks associated with four datasets (two for each task). These models are CNN and RNN, which include two models (Long Short Term Memory (LSTM)) and GRU (Gated Recurrent Unit). Each model is employed to work consequently over the two mentioned tasks to draw a road map of deep learning mod

... Show More
View Publication
Scopus (3)
Scopus Crossref
Publication Date
Thu Sep 01 2016
Journal Name
2016 8th Computer Science And Electronic Engineering (ceec)
Class-specific pre-trained sparse autoencoders for learning effective features for document classification
...Show More Authors

View Publication
Scopus (6)
Crossref (2)
Scopus Crossref
Publication Date
Tue Mar 30 2021
Journal Name
Journal Of Economics And Administrative Sciences
The Bayesian Estimation for The Shape Parameter of The Power Function Distribution (PFD-I) to Use Hyper Prior Functions
...Show More Authors

The objective of this study is to examine the properties of Bayes estimators of the shape parameter of the Power Function Distribution (PFD-I), by using two different prior distributions for the parameter θ and different loss functions that were compared with the maximum likelihood estimators. In many practical applications, we may have two different prior information about the prior distribution for the shape parameter of the Power Function Distribution, which influences the parameter estimation. So, we used two different kinds of conjugate priors of shape parameter θ of the <

... Show More
View Publication Preview PDF
Crossref
Publication Date
Sun Jul 01 2018
Journal Name
International Journal Of Engineering Research And Management
The first and Second Order Polynomial Models with Double Scalar Quantization for Image Compression
...Show More Authors

Publication Date
Thu Oct 01 2015
Journal Name
Engineering And Technology Journal
Genetic Based Optimization Models for Enhancing Multi- Document Text Summarization
...Show More Authors

View Publication
Crossref
Publication Date
Wed Dec 08 2021
Journal Name
J. Inf. Hiding Multim. Signal Process.
Predication of Most Significant Features in Medical Image by Utilized CNN and Heatmap.
...Show More Authors

The growth of developments in machine learning, the image processing methods along with availability of the medical imaging data are taking a big increase in the utilization of machine learning strategies in the medical area. The utilization of neural networks, mainly, in recent days, the convolutional neural networks (CNN), have powerful descriptors for computer added diagnosis systems. Even so, there are several issues when work with medical images in which many of medical images possess a low-quality noise-to-signal (NSR) ratio compared to scenes obtained with a digital camera, that generally qualified a confusingly low spatial resolution and tends to make the contrast between different tissues of body are very low and it difficult to co

... Show More
View Publication Preview PDF
Scopus (2)
Scopus
Publication Date
Sun Jan 01 2023
Journal Name
Journal Of Robotics And Control (jrc)
Automated Stand-alone Surgical Safety Evaluation for Laparoscopic Cholecystectomy (LC) using Convolutional Neural Network and Constrained Local Models (CNN-CLM)
...Show More Authors

In this golden age of rapid development surgeons realized that AI could contribute to healthcare in all aspects, especially in surgery. The aim of the study will incorporate the use of Convolutional Neural Network and Constrained Local Models (CNN-CLM) which can make improvement for the assessment of Laparoscopic Cholecystectomy (LC) surgery not only bring opportunities for surgery but also bring challenges on the way forward by using the edge cutting technology. The problem with the current method of surgery is the lack of safety and specific complications and problems associated with safety in each laparoscopic cholecystectomy procedure. When CLM is utilize into CNN models, it is effective at predicting time series tasks like iden

... Show More
View Publication
Scopus Crossref
Publication Date
Sat Jul 01 2017
Journal Name
Diyala Journal For Pure Science
Correlated Hierarchical Autoregressive Models Image Compression
...Show More Authors

View Publication
Crossref
Publication Date
Tue Nov 01 2022
Journal Name
2022 International Conference On Data Science And Intelligent Computing (icdsic)
An improved Bi-LSTM performance using Dt-WE for implicit aspect extraction
...Show More Authors

In aspect-based sentiment analysis ABSA, implicit aspects extraction is a fine-grained task aim for extracting the hidden aspect in the in-context meaning of the online reviews. Previous methods have shown that handcrafted rules interpolated in neural network architecture are a promising method for this task. In this work, we reduced the needs for the crafted rules that wastefully must be articulated for the new training domains or text data, instead proposing a new architecture relied on the multi-label neural learning. The key idea is to attain the semantic regularities of the explicit and implicit aspects using vectors of word embeddings and interpolate that as a front layer in the Bidirectional Long Short-Term Memory Bi-LSTM. First, we

... Show More
View Publication Preview PDF
Scopus (4)
Crossref (2)
Scopus Crossref