Abstract : DevOps is a method used to automate the process between the development team and the IT team through which they can develop, test, and release their software. Bugs during this stage slow the entire release cycle. To overcome this, Machine Learning and Deep Learning Algorithms are used to analyze and arrive at the possible cause of the bug. This reduces the dependency on the developers and, in turn, speeds up the release cycle. The bug dataset is fed to various classification algorithms like CNN, Random Forest, Decision Tree, SVM, and Na�ve Bayes for bug classification. Based on the experimental results, it can be observed that Convolutional Neural Networks, a deep learning technique, outperformed all the other approaches used. Furthermore, it was observed that Na�ve Bayes, a probabilistic classifier generally preferred for text classification, performed poorly with the bug dataset used in this paper. Ensemble methods like a Decision tree and Random Forest performed better on this dataset.

Keywords: DevOps; Machine Learning; Deep Learning; CNN; bugs

Abstract : Technology, in the year 2020, has attained its utmost use by becoming the solution for people across the world in all sectors. Despite the use of applications like Zoom, Google Meet, Kahoot, and Google Classroom, teaching language online has proved challenging without a live teacher. This paper aims to throw light on the technological gap and its pace that has not matched the necessity of the period. The analysis of an example from English Language Teaching will not show what technology lacks in this field but also serve as a guide for the developers of future artificially intelligent software and applications. As teaching and computing are two different fields, it is the researchers� duty to bridge the gap by meticulously explaining the limitations in the currently existing applications and the necessary features in the yet-to-be-developed ones.

Keywords: Insufficient Technology; ELT; applications; bridge; online teaching; Artificial Intelligence

Abstract : Controller is the heart of the distillation column, widely used in most industries such as petrochemical, pharmaceutical, and oil and gas. An intelligent and precise autonomous hybrid intelligent controller can achieve higher efficiency and high-grade pure output with the low-cost operation. The behaviour of the plant is often non-linear and interactive. Hence suitable models and perfect controller designs are very critical. A study has been carried out for several models and controllers. Extensive analysis has been carried out with different controllers and models for their efficacy, performance and the purity of the byproduct. A comparative study has been done with different controllers and models concerning their performance, and various challenges posed in numerous works of literature have been reviewed.

Keywords: Hybrid Controller; Distillation Column; artificial intelligence; low-cost operation; higher efficiency

Abstract : Hypothetically, the stability study of throughflow influence on Benard-Marangoni ferroconvection is examined. The top of the fluid layer is assumed to be free. The surface tension effect that depends on temperature is supposed to be non-deformable and subject to general thermal boundary conditions. The bottom of the fluid layer is assumed to be rigid with a fixed temperature. An analytical solution to the issue is achieved by using the Regular perturbation approach. The findings show that the stability characteristics are independent of the throughflow direction. The ferroconvection is further delayed by Peclet number Q. Convection is accelerated by raising the magnetic number Rm and the Prandtl number Pr. It is observed that the Benard-Marangoni ferroconvection is unaffected by M3, which represents the non-linearity of fluid magnetization.

Keywords:Benard-Marangoni Ferroconvection; Throughflow; Regular Perturbation Method.

Abstract : Given the online infrastructure we have today, content piracy is growing and spreading quickly across various countries. These internet infrastructures� main goal is to offer a platform for delivering permitted and lawful material from the service provider to the end user. Over time, pirates have used the digital online infrastructure system to duplicate and retransmit the original content using the same infrastructure. Visual analytics of the broadcast logo is one way to determine if the content is pirated. The construction of a new scalable TV broadcast channel logo corpus spanning diverse geographies and genres, as well as the publically available datasets of TV broadcast channel logos (Indian channels), will be covered in this work. A total of 450 TV broadcast channel logos in various regional languages have been gathered for genres including (Sports, Movies, Kids and Cartoons, Entertainment etc.) Each logo is exposed to various data augmentation approaches to increase the logo corpus and boost the deep learning logo classification. This logo corpus with the cutting-edge object identification algorithm YOLO v2 is also covered in this study, along with the recognition of several logo classes. Results from experiments are documented for various inference logos with various pixel contexts.

Keywords:Data augmentation; Deep learning; Object detection; TV broadcast; Channel logo corpus; YOLO.