Showing papers by "Shivaji University published in 2023"

Square-Facet Nanobar MOF-Derived Co₃O₄@Co/N-doped CNT Core–Shell-based Nanocomposites as Cathode Materials for High-Performance Supercapacitor Studies

Abstract: The binary as well as ternary nanocomposites of the square-facet nanobar Co-MOF-derived Co3O4@Co/N-CNTs (N-CNTs: nitrogen-doped carbon nanotubes) with Ag NPs and rGO have been synthesized via an easy wet chemical route, and their supercapacitor behavior was then studied. At a controlled pH of the precursor solution, square-facet nanobars of Co-MOF were first synthesized by the solvothermal method and then pyrolyzed under a controlled nitrogen atmosphere to get a core–shell system of Co3O4@Co/N-CNTs. In the second step, different compositions of Co3O4@Co/N-CNT core–shell structures were formed by an ex-situ method with Ag NPs and rGO moieties. Among several bare, binary, and ternary compositions tested in 6 M aqueous KOH electrolyte, a ternary nanocomposite having a 7.0:1.5:1.5 stoichiometric ratio of Co3O4@Co/N-CNT, Ag NPs, and rGO, respectively, reported the highest specific capacitance (3393.8 F g–1 at 5 mV s–1). The optimized nanocomposite showed the energy density, power density, and Coulombic efficiency of 74.1 W h.kg–1, 443.7 W.kg–1, and 101.3%, respectively, with excellent electrochemical stability. After testing an asymmetrical supercapacitor with a Co3O4@Co/N-CNT/Ag NPs/rGO/nickel foam cathode and an activated carbon/nickel foam anode, it showed 4.9 W h.kg–1 of energy density and 5000.0 W.kg–1 of power density.

Machine Learning for Detecting Noisy Neighbours in 5G Networks

Abstract: The structure of 5G networks is predicted to become more dynamic as well as unpredictable than that of present networks. Network Functions (NF) will cease to exist firmly tied through the hardware on which they run with the arrival of Network Function Virtualization (NFV), posing new issues in network administration. A noisy neighbour is a term widely used in NFV architecture to describe circumstances in which an application's performance suffers because a few of the assets it requires are consumed through other apps on the same cloud node. Such circumstances are difficult to identify using simple procedures, necessitating the employment of complex methodologies to NFV infrastructure management. We show how Machine Learning (ML) approaches may be utilised to sense such occurrences in this work. We demonstrate that conventional models intended for automatic classification can sense the noisy neighbour phenomena with greater than 90% correctness in a basic scenario using data gathered at actual NFV infrastructure.

Detecting Multi-Pose Masked Face using Haar Cascade and LBP Classifiers

Abstract: This study presents the Haar cascade and Local Binary Pattern algorithms to detect faces. Facial detection technology can identify a person's face in a pictures or videos. The significance of face detection has increased as modern cultures have developed, primarily because of the necessity for security, including legal requirements, global security, and other pressing situations. This research compares the performance of Haar Cascade and Local Binary Pattern in a real-time system for human face detection. The face detection technique uses both frontal and profile face detectors in a cascade of boosted classifiers to extract the “Haar” traits and apply them. The method integrates profile and frontal face cascades and categorizes the face based on the position by comparing it to a group of faces with comparable ranges. Algorithms are tested on different images of single pose, multi-pose and masked faces. The results show that the Haar cascade classifier outperforms the LBP classifier.

Restaurant Quality Analysis: A Machine Learning Approach

Abstract: Managing customer’s happiness has emerged as a significant business trend, particularly in the restaurant industry. The purpose of this study is to determine how K-Means algorithms can be used to measure customer satisfaction at a family restaurant in Kolhapur. A survey is carried out related to services and ambiguous at the restaurant. What makes restaurants popular is the main focus of the survey. Data collected through online survey are clustered using the elbow method as well as the K-Means clustering. This study presents the results of the customer satisfaction measurement and offers improvement and recommendations to the concerned restaurant.

Machine Learning Assisted Analysis, Prediction, and Fabrication of High‐Efficiency CZTSSe Thin Film Solar Cells

Abstract: The Earth-abundant element-based Cu2ZnSn(S,Se)4 (CZTSSe) absorber is considered as a promising material for thin-film solar cells (TFSCs). The current record power conversion efficiency (PCE) of CZTSSe TFSCs is ≈13%, and it's still lower than CdTe and CIGS-based TFSCs. A further breakthrough in its PCE mainly relies on deep insights into the various device fabrication conditions; accordingly, the experimental–oriented machine learning (ML) approach can be an effective way to discover key governing factors in improving PCE. The present work aims to identify the key governing factors throughout the device fabrication processes and apply them to break the saturated PCE for CZTSSe TFSCs. For realization, over 25,000 data points were broadly collected by fabricating more than 1300 CZTSSe TFSC devices and analyzed them using various ML techniques. Through extensive ML analysis, the i-ZnO thickness is found to be the first, while Zn/Sn compositional ratio and sulfo-selenization temperature are other key governing factors under thin or thick i-ZnO thickness to achieve over 11% PCE. Based on these key governing factors, the applied random forest ML prediction model for PCE showed Adj. R2 = >0.96. Finally, the best-predicted ML conditions considered for experimental validation showed well-matched experimental outcomes with different ML models.

Secure Authentication of IoT Devices Using Upgradable Smart Contract and Fog-Blockchain Technology

Abstract: There are many new challenges related to the security of the Internet of Things networks due to the rapid growth of the technology. One of the main challenges is device security and authentication. Due to their limited processing and storage capabilities, the Internet of Things devices are unable to protect themselves from different attacks. Passwords or predefined keys have drawbacks that limit their uses. With high trust, integrity, and transparency, distributed ledger technology has the ability to address these issues in IoT. The solution discussed in this paper is to use fog computing and blockchain technology to authenticate IoT devices using smart tokens generated by an upgradable smart contract.

Measuring Regional Dimensions of Sustainable Livelihood Security Index in Maharashtra

Abstract: Abstract The Sustainable Livelihood Security Index (SLSI) is an integrated effective and advantageous tool to measure agriculture sustainability and livelihood security. The sustainable Livelihood Security Index mechanism can help to track and estimate the agricultural sector's growth and development status. This study deals with to measuring of sustainability of the agriculture sector through the Sustainable Livelihood Security Index. SLSI approach is inclusive tool in terms of ecological, economic, and social aspects and its various sub indicators, which significantly impact the achievement of the Sustainable Development Goal (SDG). The main objective of this study, is to measure regional development of Maharashtra during two time periods, i.e., 2010 and 2019. Finding of study shows that western Maharashtra region is ranked highest and Vidharbha region is least developed in SLSI among all five regions in Maharashtra. The SLSI value of western Maharashtra has improved from 0.525 to 0.546 during 2010-19. However, the Vidharbha region though had lesser SLSI value as compared to other regions of Maharashtra moreover there is improvement in SLSI from 0.289 to 0.330 during same period. Overall SLSI value of Maharashtra state has raised from 0.404 to 0.414 during 2010-19. This reveals that there is positive net change in SLSI to the extent of 2.6% during 2010-19.

A Survey on Controllable Abstractive Text Summarization

Abstract: The main objective of automatic text summarization is to create a clear and precise summary while maintaining the essential information content and overall meaning. For the purpose of automatically summarising material, multiple techniques have recently been created and are now widely used across numerous fields. The two main strategies for automatic summarization are extraction and abstraction. While abstractive summarization approaches seek to produce relevant information in a novel way, extractive summarization techniques function by recognising important chunks of the text from documents. Abstractive text summarization (ATS) is one of the popular areas of research. Numerous models have been created for ATS. However, the generic summaries produced by the current summarising models do not take into account the preferences or expectations of the consumers. These systems ignore the preferences of the user, including the preferred length, style, entities, or how much of the information has already been read by the user, in which they may be interested. The ability to limit the summary length in abstractive summarization is a significant problem because generated summaries are employed in various scenarios that may have space or length constraints. Document summarization models, like people, have a variety of methods to understand a document's contents. We look into controllable text summarization that gives users some degree of control over a certain aspect of the output. The work that has been done in the field of controlled abstractive summarization is reviewed in this study. Additionally, we have provided a summary of the many controlled abstractive summarization approaches that have been utilised or developed by researchers as well as the assessment methodologies that are currently being used to these summaries.

Video Event Description Booster By Bi-Modal Transformer, Identity, and Emotions Capturing

Abstract: In today’s era, interactive media material consisting of medium text, audio, picture, and video basically has a multimedia style. With the assistance of advanced deep learning (DL) methods, some of the rare computer vision (CV) issues have been successfully resolved. Video Captioning is automatic description generation from digital video. It converts the audio tracks of people within a video into text. When the recording is played that text will be displayed in segments that are synchronized with specific words as they are spoken. It is like representing a whole activity which is happening in a video in a textual format. The captions which are generated in the process of video captioning are nothing but a transcription of dialogue and visual content in a video. They appear as a text on the bottom of the display screen [1, 2]. In this paper, we have studied and modified video captions generated using Bi-modal Transformer (BMT). We have also added facial recognition features to BMT to give more information about the expressive actions in the video. The proposed model also detects emotional expressions on a person’s face in a video.

Multifunctional Fluorescent Tetraphenylethene-Based Reversible Mechanochromism for Highly Selective Detection of MnO₄^– in Aqueous Media and Green Organic Light Emitting Diode Applications

Abstract: Ethyl 2-cyano-3-(4-(1,2,2-triphenylvinyl) phenyl) acrylate (TPE-SKJ), a newly synthesized luminogen based on tetraphenylethene, with single crystal analysis exhibits photophysical phenomena such as aggregation-induced emission (AIE); reversible mechanochromic, solvatochromic, organic light emitting diode; and chemical sensing in aqueous media with great selectivity and a low limit of detection. The synthesized material demonstrates high selectivity and sensitivity capacity for sensing MnO4– in mixed aqueous media (water/acetonitrile, v/v, 9/1). The detection limit for MnO4– was found to be 0.086009 μg mL–1 with a quantum yield (Φ) of 11%. Moreover, we employed TPE-SKJ material in an organic light-emitting diode (OLED) as an emissive layer. The device shows a maximum of 1.62% external quantum efficiency, higher than nondoped emitting layer-based green OLEDs. The present results will encourage ongoing research into the design of novel stimuli-responsive organic materials with switchable properties based on their supramolecular interactions for numerous applications.

Recent Advances in Magnetorheological Damping Systems for Automotive Technology

Abstract: Magnetorheological (MR) fluids are smart fluids that respond to an external magnetic field. Numerous companies are continuously researching faster response times and intelligent suspension automotive technology. A significant growth rate of a smart MR damper for automotive technology is examined in this article, along with the value chain and market trends. The commercial automobile MR damper is elaborate, and their MR technology in automotive suspension systems is comprehended. The current state of technology and research are discussed in terms of future market development.

Secure QR Code Transactions Using Mobile Banking App

Abstract: Mobile banking applications have revolutionized the manner in which financial transactions are executed and have made the customer's life easy by handling financial transactions which have improved the quality of customer service in the banking sector. During the COVID-19 pandemic situation, financial transactions employing mobile banking apps were at their peak, gained tremendous importance, and became more popular. The current paper reviews popular and most frequently used mobile banking services provided by public and private sector banks. The authors have designed and developed a model for a secure mobile app for performing financial transactions securely by employing a QR code which provides a low-cost solution to the problem under consideration. The CIA (Confidentiality, Integrity, and Authenticity) triad is taken care of in the model implementation employing existing security techniques such as hashing, role-based authentication mechanism, prevention of SQL injection attacks, and tracking MAC address of a user. The possibilities of QR code hacks are presented and the solutions are proposed. The double spending problem is tracked.

Driving Style Classification Using Deep Learning Techniques

Abstract: The increased interest in driving analysis is a result of growing worries about car safety. However, the lack of publicly accessible driving data currently hinders the development in this area. Since machine learning approaches require enormous volumes of data that are expensive and difficult to collect through Naturalistic Driving Studies, they are only partially accessible to the public research community. A cheap and simple platform for driver behavior detection has also been made available by the rise of smart phones; however, the data from these applications is not publicly accessible. Due to these factors, this study provides an available dataset UAH-Drive Set that enables deep driving. This dataset is an imbalanced dataset and a sampling technique must be used to make it a balanced dataset. By supplying a significant amount of data gathered by the system, the dataset is sampled first and after that, the machine learning algorithms are applied. The programmed system is handled by three distinct behavior, three different drivers and cars (slow, aggressive, and normal) on two different types of roads (a side road and a highway), resulting in more than 8 h of realistic driving along with the associated unprocessed data, the semantic data as well as the video recordings of the trips. Additionally, a tool that aids in plotting the display, the travel videos, and data are concurrent in order to easily analyze the data.