Vellore Institute of Technology (VIT) is a private university located in&nbsp;Tamil Nadu, India. Founded in 1984, as Vellore Engineering College, the institution offers 20 undergraduate, 34 postgraduate, four integrated and four research programs. It has campuses in Vellore, Amravati, Bhopal and Chennai.

VIT is one of the top ranked private universities in India according to NIRF, THE and QS Rankings.&nbsp;Govt. of India has recognized&nbsp;VIT, Vellore as an&nbsp;Institution of Eminence. This has allowed VIT to take independent quality initiatives and move up in world ranking.

&nbsp;

&nbsp;

VIT University

Department of Software Systems

Pneumonia can be described as a disease caused via viruses, bacteria, and fungi which lead to the inflammation of the lung and decreased oxygen absorption from air. It is one of the leading causes of death globally, reaching up-to 100,000 deaths caused by it globally. It is historically been challenging to identify this illness just by looking at chest X-rays. This study aims to smoothen the process of efficiently and accurately detecting pneumonia using Image Recognition based deep learning algorithms. This chapter aims to solve this problem using novel machine learning frameworks involving the Efficient-Net and InceptionV3 algorithms to achieve this goal. Our proposed models have achieved an accuracy of 92.93% via the InceptionV3 model and 95.39% via the EfficientNet model.

School of Computer Science and Engineering

Vellore Campus

Deep Learning Applications in Image Analysis

Chest X-Ray Image Classification of Pneumonia Disease Using EfficientNet and InceptionV3

Farmers face several problems due to diseases that affect their plants. To diagnose the same, well-equipped lab is required and it is not efficient solution and not even budget-friendly. To provide all the farmers and cultivators with smartphones with internet access, we could reduce the food loss in the country. In this chapter we have covered how deep learning can be used to make an image classifier based on AlexNet. The trained weight is then deployed to classify new images according to diseases present on leaves. To train the model, we have used the dataset that contains 38 classes of different diseases. As a result, we are able to achieve 91.3% accuracy on test set using AlexNet model

Plant Diseases Classification Using Neural Network: AlexNet

Reviews exhaustively the key recent research into deep learning applications in image analysisCovers many different deep-learning applications in medical, satellite, forensic image analysisDemonstrates the deep learning approach as effective solutions for various image-related problems

<a href="https://scholar.google.com/citations?view_op=view_citation&amp;hl=en&amp;user=fnMkkMkAAAAJ&amp;sortby=pubdate&amp;citation_for_view=fnMkkMkAAAAJ:QUX0mv85b1cC">Deep Learning Applications in Image Analysis</a>

Natural Language Generation (NLG) is one of the most critical yet challenging tasks in all Natural Language Processing applications. It is a process to automate text generation so that humans can understand its meaning. A handful of research articles published in the literature have described how NLG can produce understandable texts in various languages. The use of sequence-to-sequence modeling powered by deep learning techniques such as Long Term Short Term Memory, Recurrent Neural Networks, and Gated Recurrent Units has received much popularity as text generators. This survey provides a comprehensive overview of text generations and their related techniques, such as statistical, traditional, and neural network-based techniques. Generating text using the sequence-to-sequence model is not a simple task as it needs to handle continuous data, such as images, and discrete information, such as text. Therefore, in this study, we have identified some crucial areas for further research on text generation, such as incorporating a large text dataset, identifying and resolving grammatical errors, and generating extensive sentences or paragraphs. This work has also presented a detailed overview of the activation functions used in deep learning-based models and the evaluation metrics used for text generation.

World University root node

Neural Processing Letters

Natural Language Generation Using Sequential Models: A Survey

It is undeniable that social media has improved our lives in many ways, like allowing interactions with others all over the world and network expansion for businesses. However, there are detrimental effects of such accessibility, including the rapid spread of hate through offensive messages typically directed toward gender, religion, race, and disability, which can cause psychological harm. To address this problem of social media, many researchers have recently proposed various algorithms powered by machine learning (ML) and deep learning for the detection of hate speech. This work proposes a hate speech detection model based on long-short term memory (LSTM), using term frequency inverse document frequency (TF-IDF) vectorization, and makes comparisons with support vector machine (SVM), Naïve Bayes (NB), logistic regression (LR), XGBoost (XGB), random forest (RF), K -nearest neighbor ( k -NN), artificial neural network (ANN), and bidirectional encoder representations from transformers (BERT) models. To validate and authenticate our proposed work, we obtained and classified a real-time Twitter data stream of a trending topic using Twitter API into two classes: hate speech and nonhate speech. The precision, recall, and F 1 score achieved by LSTM are 0.98, 0.99, and 0.98, respectively. The accuracy of LSTM for detecting hateful sentiment was found to be 97%, surpassing the accuracy of other models.

IEEE Transactions on Computational Social Systems

Hateful Sentiment Detection in Real-Time Tweets: An LSTM-Based Comparative Approach

Foundation settlement is an important design criterion as it affects the durability of a structure. Conventional methodologies calculate only a global factor of safety to determine the safety of the structure. However this does not account for the uncertainties due to soil variability and measurement errors. Therefore reliability based design principles must be incorporated to determine the performance and reliability of a structure. The first order second moment method (FOSM) is generally used for this analysis but it is time consuming. On the other hand, relevance vector machine (RVM) achieves very good generalisation performance. Thus in our study we have used RVM-based FOSM and ELM and compared the results obtained from both. For this, a dataset of 480 readings was developed for cohesive frictional soil taking Poissons ratio and elastic modulus parameters as random variables. Readings used for training was 70\% and 30\% were used for testing. Normalised data was used. Additionally, several error and correlation functions were also calculated to assess the performance of the models.

International Journal of Advanced Intelligence Paradigms

Reliability analysis of shallow foundation based on settlement criteria

CMES-Computer Modeling in Engineering \& Sciences

Vessels segmentation in angiograms using convolutional neural network: A deep learning based approach

River flow modeling is essential for critical aspects such as effective water management and structure planning, together with flood and drought analysis. There has been a growing interest in modeling hydrological systems via machine learning (ML) models. Various optimization techniques are utilized to develop the applications of ML-based hydrological models. The ultimate aim of this research is to establish high performance forecasting model. Therefore, this study conducts river flow modeling by using the daily data attained from a gauge station situated in the Euphrates Basin. For this purpose,&nbsp;Artificial Neural Network (ANN) model was hybridized with five different optimization algorithms i.e.,&nbsp;Artificial Bee Colony (ABC), Teaching-Learning Based Optimization (TLBO), Ant Colony Optimization (ACO), Ant-Lion Optimization (ALO), and Imperialist Competitive Algorithm (ICA). In determining the inputs used to create the models, the distribution graph and correlation of the data with the previous period data were examined. The results were evaluated with eleven different statistical parameters and an error matrix. Examining the obtained results, the study revealed all models present high performance. When the results were reviewed in general, it was seen that all determination coefficient (R2) and Nash-Sutcliffe coefficient (NSE) values were higher than 0.962, and other parameters were very close to the optimum. By comparing all the developed hybrid models, ANN-ALO model reported the highest performance for river flow forecasting.

Earth Science Informatics

Comparison of different optimized machine learning algorithms for daily river flow forecasting

first_pagesettings<a href="https://www.mdpi.com/1999-5903/14/11/340/reprints">Order Article Reprints</a>Open AccessArticle<h2>Multimodel Phishing URL Detection Using LSTM, Bidirectional LSTM, and GRU Models</h2>by<figure class="image"><img src="https://www.mdpi.com/bundles/mdpisciprofileslink/img/unknown-user.png"></figure><a href="https://sciprofiles.com/profile/707924">Sanjiban Sekhar Roy</a>1<figure class="image"><img src="https://pub.mdpi-res.com/img/design/orcid.png?0465bc3812adeb52?1689945757"></figure>,<figure class="image"><img src="https://www.mdpi.com/profiles/275364/thumb/Ali.Awad.jpg"></figure><a href="https://sciprofiles.com/profile/275364">Ali Ismail Awad</a>2,*<figure class="image"><img src="https://pub.mdpi-res.com/img/design/orcid.png?0465bc3812adeb52?1689945757"></figure>,<figure class="image"><img src="https://www.mdpi.com/bundles/mdpisciprofileslink/img/unknown-user.png"></figure><a href="https://sciprofiles.com/profile/2547783">Lamesgen Adugnaw Amare</a>1,<figure class="image"><img src="https://www.mdpi.com/bundles/mdpisciprofileslink/img/unknown-user.png"></figure><a href="https://sciprofiles.com/profile/2600472">Mabrie Tesfaye Erkihun</a>1 andMohd Anas1<figure class="image"><img src="https://pub.mdpi-res.com/img/design/orcid.png?0465bc3812adeb52?1689945757"></figure>1School of Computer Science and Engineering, Vellore Institute of Technology, Vellore 632014, India2College of Information Technology, United Arab Emirates University, Al Ain P.O. Box 17551, United Arab Emirates*Author to whom correspondence should be addressed.Future Internet 2022, 14(11), 340; <a href="https://doi.org/10.3390/fi14110340">https://doi.org/10.3390/fi14110340</a>Received: 12 September 2022 / Revised: 17 November 2022 / Accepted: 18 November 2022 / Published: 21 November 2022&nbsp;(This article belongs to the Special Issue <a href="https://www.mdpi.com/journal/futureinternet/special_issues/CCASM">Cybersecurity and Cybercrime in the Age of Social Media</a>) &nbsp;Download<a href="https://www.mdpi.com/1999-5903/14/11/340#">Browse Figures</a>&nbsp;<a href="https://www.mdpi.com/1999-5903/14/11/340/notes">Versions&nbsp;Notes</a>&nbsp;<h2>Abstract</h2>In today’s world, phishing attacks are gradually increasing, resulting in individuals losing valuables, assets, personal information, etc., to unauthorized parties. In phishing, attackers craft malicious websites disguised as well-known, legitimate sites and send them to individuals to steal personal information and other related private details. Therefore, an efficient and accurate method is required to determine whether a website is malicious. Numerous methods have been proposed for detecting malicious uniform resource locators (URLs) using deep learning, machine learning, and other approaches. In this study, we have used malicious and benign URLs datasets and have proposed a detection mechanism for detecting malicious URLs using recurrent neural network models such as long short-term memory (LSTM), bidirectional long short-term memory (Bi-LSTM), and the gated recurrent unit (GRU). Experimental results have shown that the proposed mechanism achieved an accuracy of 97.0% for LSTM, 99.0% for Bi-LSTM, and 97.5% for GRU, respectively.

Fulltext

Future Internet

Multimodel Phishing URL Detection Using LSTM, Bidirectional LSTM, and GRU Models

dentification of histone modification from datasets that contain high-throughput sequencing data is difficult. Although multiple methods have been developed to identify histone modification, most of these methods are not specific for histone modification but are general methods that aim to identify protein binding to the genome. In this study, tensor decomposition (TD) and principal component analysis (PCA)-based unsupervised feature extraction with optimized standard deviation were successfully applied to gene expression and DNA methylation. The proposed method was used to identify histone modification. Histone modification along the genome is binned within the region of length L. Considering principal components (PCs) or singular value vectors (SVVs) that TD or PCA attributes to samples, we can select PCs or SVVs attributed to regions. The selected PCs and SVVs further attribute P-values to regions, and adjusted P-values are used to select regions. The proposed method identified various histone modifications successfully and outperformed various state-of-the-art methods. This method is expected to serve as a de facto standard method to identify histone modification.

bioRxiv

Tensor decomposition and principal component analysis-based unsupervised feature extraction outperforms state-of-the-art methods when applied to histone modification profiles