UNICEF reports that health workers frequently use the child’s breath count to diagnose pneumonia in nations with weak health care systems (i.e., few doctors, limited access to laboratories, and chest x-rays). Counting a child’s respiratory rate (RR) manually using an acute respiratory infection timer is a challenging task among health workers. The misclassification of an observed rate would lead to improper treatment. In the work, a low-cost noncontact automatic counter was proposed to estimate the RR using a Time of Flight (ToF) sensor. Here, the ToF sensor is pointed at the chest of the human body with clothes to capture chest wall expansion and contraction. This chest wall movement can be related to the inhalation and exhalation of breathing, which can be related to respiration activity. Clear breathing patterns were observed in the extracted respiration signal and standard signal processing techniques were employed to estimate the respiration rate. The device can measure the subject’s respiration rate at a wide range of frequencies (slow, normal, and rapid), ranging from 4 to 100 breaths per minute (bpm). The single-pixel ToF sensor with a 940 nm wavelength is illumination-invariant in indoor applications and robust against clothes. The device’s performance was assessed by comparing the attained real-time signals with the ground truth signals. The proposed device’s estimated mean absolute error (MAE) has an error of less than 2 bpm for any frequency range. The device is computationally efficient, taking less than 36 ms to estimate and display the RR readings in the user interface.

Publishing highlights after a sports game is a common practice in the broadcast industry, providing viewers with a quick summary of the game and highlighting interesting events. However, the manual process of compiling all the clips into a single video can be time-consuming and cumbersome for video editors. Therefore, the development of an artificial intelligence (AI) model for sports highlight generation would significantly reduce the time and effort required to create these videos and improve the overall efficiency and accuracy of the process. This would benefit not only the broadcast industry but also sports fans who are looking for a quick and engaging way to catch up on the latest games. The objective of the paper is to develop an AI model that automates the process of sports highlight generation by taking a match video as input and returning the highlights of the game. The approach involves creating a list of words (wordnet) that indicate a highlight and comparing it with the commentary audio’s transcript to find a similarity, making use of a speech-to-text conversion, followed by some pre-processing of the extracted text, vectorization and finally measurement of the cosine similarity metric between the text and the wordnet. However, this process can become time-consuming too, in case of longer match videos, as the computation times of the AI models become inefficient. So, we used a parallel processing technique to counter the time required by the AI models to compute the outputs on large match videos, which can decrease the overall time complexity and increase the overall throughput of the model.

Scientific studies of the elements that influence the box office performance of Indian films have generally concentrated on post-production elements, such as those discovered after a film has been completed or released, and notably for Bollywood films. Only fewer studies have looked at regional film industries and pre-production factors, which are elements that are known before a decision to greenlight a film is made. This study looked at Indian films using natural language processing and machine learning approaches to see if they would be profitable in the pre-production stage. We extract movie data and English subtitles (as an approximation to the screenplay) for the top five Indian regional film industries: Bollywood, Kollywood, Tollywood, Mollywood, and Sandalwood, as they make up a major portion of the Indian film industry’s revenue. Subtitle Vector (Sub2Vec), a Paragraph Vector model trained on English subtitles, was used to embed subtitle text into 50 and 100 dimensions. The proposed approach followed a two-stage pipeline. In the first stage, Return on Investment (ROI) was calculated using aggregated subtitle embeddings and associated movie data. Classification models used the ROI calculated in the first step to predicting a film’s verdict in the second step. The optimal regressor–classifier pair was determined by evaluating classification models using [Formula: see text]-score and Cohen’s Kappa scores on various hyperparameters. When compared to benchmark methods, our proposed methodology forecasts box office success more accurately.

A new array P system variant, parallel 8-directional array P systems, is introduced in this paper. In addition, we have found an application for such a system when moving a robot in a rectangular grid from source to destination without colliding with obstacles. The following four cases of robot motion have been discussed. In the grid, we start with a fixed number of static obstacles, followed by a fixed number of dynamic obstacles moving with constant speed, a fixed number of dynamic obstacles moving with varying speeds, and finally, varying numbers of dynamic obstacles moving with constant speed. All these algorithms were solved in polynomial time.

Amidst the profound impact of the COVID-19 pandemic on global economies and healthcare systems, effective data analysis has become paramount. Our research paper, titled ”Data Analytics for Pandemic Management Using MapReduce and Apriori Algo- rithm,” presents a comprehensive framework to analyze pandemic data. We harness the power of the MapReduce and Apriori algorithms, with a parallel processing model achieving an average speedup of 50%. This approach involves data collection, pre- processing, and algorithmic application to extract valuable insights from pandemic-related data. Notably, our findings reveal a substantial 22.29% support rate for ”n95 masks,” indicating high demand. Additionally, we identify strong co-occurrence patterns, exemplified by the perfect support rate of 1.00 between ”n95 masks” and ”chloroquine,” highlighting their interconnectedness. Our framework goes beyond data analysis, enhancing personalized marketing, and optimizing inventory management for efficient resource allocation during crises. We also uncover robust associations, such as the 0.220 confidence in joint purchases of ”butyl rubber gloves” and ”surgical masks”.Hence, the research aligns seamlessly with the pressing need for effective pandemic data management and response strategies. By validating our approach through numerical insights, we aim to contribute to mitigating pandemic-related challenges and support global efforts to better prepare for and manage future health emergencies.

The recognition of gestures with accuracy is still a widespread challenge which leads to prevention in the usage of wearable gesture recognition systems. In this paper, we proposed a gesture-based mouse, a smart glove based system for gesture recognition and remote control. To reduce the dependency of the gesture recognition system on external conditions like light and network connection, we eliminate the need of a camera and use a glove fitted with the MPU6050 sensor to detect motion accurately with minimal latency. The elimination of the camera component also reduced the overall component cost significantly. In this paper, we discuss the algorithm followed to map hand motion to cursor movement on the screen via the pseudocode and flowchart. Using this methodology, basic functionalities of the mouse and its operations were implemented with less expense and to more effectiveness.

This paper proposes a method for generating and solving mazes using minimum spanning tree algorithms and parallelizing them using OpenMP. The approach involves generating a random maze using Prim's and Kruskal's algorithms, which constructs a minimum-spanning tree from a set of edges. The maze is then solved using Dijkstra's algorithm, which finds the minimum spanning tree from a given starting point. To optimize the algorithms' performance, the parallel processing capabilities of OpenMP are utilized to distribute the workload across multiple processors. The results of the experiments demonstrate the effectiveness of the proposed method in generating and solving mazes efficiently, with significant reductions in computation time achieved through parallelization.

Bitcoin is the world’s most traded cryptocurrency and highly popular among cryptocurrency investors and miners. However, its volatility makes it a risky investment, which leads to the need for accurate and fast price-prediction models. This article proposes a Bitcoin price-prediction model using a long short-term memory (LSTM) network in a distributed environment. A tensor processing unit (TPU) has been used to provide the distributed environment for the model. The results show that the TPU-based model performed significantly better than a conventional CPU-based model.

In today’s digital world, most organizations are prone to cyberattacks. As a result, they face huge data and economic loss. Even under some circumstances, the organizations could lose their reputations and identity. A lot of research was conducted to address these cyber-attacks but still, it is a huge threat. Most of the algorithms address only the entry-level attacks and fail to replicate that performance to other attacks. Intrusion detection is one serious issue that can destabilize any kind of network. Especially, they are very difficult to contain in real-time systems. If these attacks are not detected in the early stages, they can create serious consequences for the network. The objective of the paper is to create a system that uses various machine learning algorithms to classify and predict network intrusions. Learning models are created and applied to large databases. Finally, these models were tested using various evaluation indicators for real-time data, and results are compared under various scenarios and use cases.

Nowadays, microcontrollers are very common in a wide range of applications. Integrating Machine Learning with Embedded Systems is essential and should be provided at a low cost to satisfy modern-day requirements. The tools that currently exist for this purpose can be used only with the hardware provided by the same vendor, and are usually not flexible with IDEs. This paper proposes an open-source framework that helps to deploy a trained Neural Network (Both Convolution and Deep Neural Network) on any ARM Cortex M-based microcontroller, providing an easy and low-cost device. It converts the trained model with float32 parameters into int8 type parameters and automatically generates inference code using SIMD optimized CMSIS functions. The output obtained using the proposed framework is validated with the X-Cube-AI framework. Also, A voice-based digit classification system is implemented using a Cortex-M4 microcontroller to demonstrate the efficiency of this procedure.

Solving the system of linear systems is of paramount importance in the field of science and technology. The applications of linear equations have been widely found in diverse fields. As an effect of the massive rise of big data, developing computational algorithms for solving systems of linear equations of large size has gained utmost importance in the field of data science. Since solving large systems of linear equations serially can be time consuming and slow process, parallelizing the algorithms with appropriate parallel constructs provides the accurate solutions with less time complexity. Parallel implementation of three algorithms namely back substitution, conjugate gradient and Gauss Seidel to solve large systems of linear equations using OpenMP is proposed in this paper. To determine the most time and space efficient method among the three, comparative analysis of both the serial and parallel execution of each algorithm are presented. The algorithms are further optimized to get the best results for the execution time. The algorithms for solving the system of linear equations generally involve multiple steps. So, by scheduling the number of threads for each process the execution time of the process could be optimized. While executing the programs parallelly different number of threads run simultaneously on different processors. These aspects have been considered to propose the implementation of the three algorithms. The limitations of particular algorithms are considered, their laws of convergence are discussed and comparative analysis is developed considering those kinds of matrices which can be solved accurately by all the three algorithms. A complete comparative study of all the three algorithms computed parallelly and serially with detailed analysis of the performance are presented.

A file system is a collection of rules that govern how files are labelled, maintained, and accessed from a storage medium. Initially, there were a lot of different file systems within different servers and machines. Various file operations are allowed only between the files present in the same operating systems. When we have files in different operating systems these operations cannot be performed as the file system does not allow it. The virtual file system becomes an abstract overlay over a more tangible file system, which allows heterogeneous file transfer among different Operating Systems. Although the features offered by both the File Systems are similar, the Virtual File System provides us with an environment, wherein the Files can be accessed by any Operating System type and File System type. This paper focuses on comparing the time it takes to complete the different file operations like creating, reading, writing, deleting a file, etc. using the VFS and a traditional FS. We believe this study would help in better understanding the benefits of using a VFS.

Vehicular ad hoc network (VANET) is a modern communication design that allows the interpretation between vehicles to move at high speeds on the roads. It has started the opportunity to design various new administrations like traffic administration, distribution of emergency data to avoid dangerous conditions, traffic engineering, and other user records. VANETs are a close relation of MANETs but with recognizing features like action at high speeds, processing power, moving controlled, sufficient storage, variable node consistency, and challenging communication position with precise link existence. Various security issues are usually identified in vehicular ad hoc networks, like authentication and authorization issues, and it gets problems from frequent attacks that survive for MANET. As VANETs are extensively used in various critical applications, much investigation has been done to determine effective solutions against these DoS attacks that can work for multiple routing protocols. This paper presents a brief survey on DoS attack on network availability and its extremity stage in the VANET atmosphere. Also, the paper discusses on various existing techniques to detect and prevent the DoS attack. A detailed comparative statement is made between these techniques to understand its effectiveness in VANET.

The high-performance computing paradigm needs high-speed switching fabrics to meet the heavy traffic generated by their applications. These switching fabrics are efficiently driven by the deployed scheduling algorithms. In this paper, we proposed two scheduling algorithms for input queued switches whose operations are based on ranking procedures. At first, we proposed a Simple 2-Bit (S2B) scheme which uses binary ranking procedure and queue size for scheduling the packets. Here, the Virtual Output Queue (VOQ) set with maximum number of empty queues receives higher rank than other VOQ's. Through simulation, we showed S2B has better throughput performance than Highest Ranking First (HRF) arbitration under uniform, and non-uniform traffic patterns. To further improve the throughput-delay performance, an Enhanced 2-Bit (E2B) approach is proposed. This approach adopts an integer representation for rank, which is the number of empty queues in a VOQ set. The simulation result shows E2B outperforms S2B and HRF scheduling algorithms with maximum throughput-delay performance. Furthermore, the algorithms are simulated under hotspot traffic and E2B proves to be more efficient.

Merkle-Hellman public key cryptosystem is a long-age old algorithm used in cryptography. Despite being computationally fast, for very large input sizes it may operate slower due to thread creation overhead or reaching a deadlock situation. In this paper, we discuss the working principles of the Traditional Merkle-Hellman knapsack cryptosystem, which is an Easy knapsack. The challenges of Hard Knapsack and how it overcomes the shortcomings of the Traditional Easy Knapsack, are also discussed. The Hard knapsack variant of Merkle-Hellman is solved first using plain recursion and then improvised using a dynamic programming approach to the problem. Parallelism and Concurrency has been achieved on the dynamic programming implementation using OpenMP API which further has enhanced the performance time. A comparative study of both variants of Hard Knapsack for messages of different lengths has shown that the latter is faster.

Wearable sensors for activity recognition can give a response to the person about their daily life concerning sports physical activity. Moreover, it encourages a more active lifestyle. Mostly, supervised laboratory settings are used to study activity recognition. Therefore, this paper gives a comparative analysis of different types of classification techniques for classifying data obtained from body-worn magnetic and inertial sensors. The algorithms used for classification and comparison are Decision Tree (DT), Gradient Boosting Classification, Bagging Classification, Random Forest Classifier, Extra Trees Classifier, K-Nearest Neighbours (K-NN), Support Vector Machines (SVM), Neural Networks and its variants. The daily and sports activity are recorded using sensors worn on the right hand, left hand, torso, right leg, left leg. Each sensor consists of a triaxial gyroscope, triaxial accelerometer, triaxial magnetometer. Features extracted from raw sensor data using Singular Value Decomposition (SVD) and Principal Component Analysis (PCA) is used for classification.

Deep Neural Networks have been tremendously successful in a wide variety of applications, especially those that hold high importance to the society. It has thus ignited a demand among researchers and data scientists to build larger Deep Neural Networks. The problem with these kinds of Neural Networks is that a huge amount of data is needed to train the model. Hence the computational and processing recourses become a bottleneck after a certain limit. The learning problems in the field of structural chemistry and bio-chemistry involve classification of proteins sequences as a fundamental step. Sequential training of these models would be computationally rigorous and time consuming. This invites the notion of distributed training and employing multi-core architectures to train such Neural Networks. Again, the solution is not as simple as distributing the data over multiple machines for training. The Stochastic Gradient Descent algorithm is sequential by nature, there is a need for asynchronous implementation for avoiding various bottlenecks. Hence, this paper trains an Artificial Neural Network by distributing the dataset of protein sequences over multiple GPUs by using mini-batch Asynchronous Stochastic Gradient Descent Algorithm. The Neural Network trained under distributed environment reduces the training time significantly compared to its sequential implementation.

Floods often turn out to be a major natural disaster in some parts of world due to overflow of water which submerges land that is usually dry. This leads to loss of life and vast damage to economy; therefore, it becomes extremely important to have systematic and dynamic prediction of flood areas so that people are more aware and better prepared for the impending disaster. The aim of the paper is to develop a comprehensive model for the prediction of flood area using Apriori algorithm. Our primary focus is to assess the spatial dataset of Bangladesh which provides hazard data of various risks including the risk of flooding. The various factors resulting in flooding such as water level and flood area have been analysed and the possible relations have been developed using the association rules.

This article has been withdrawn: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/our-business/policies/article-withdrawal). This article has been withdrawn as part of the withdrawal of the Proceedings of the International Conference on Emerging Trends in Materials Science, Technology and Engineering (ICMSTE2K21). Subsequent to acceptance of these Proceedings papers by the responsible Guest Editors, Dr S. Sakthivel, Dr S. Karthikeyan and Dr I. A. Palani, several serious concerns arose regarding the integrity and veracity of the conference organisation and peer-review process. After a thorough investigation, the peer-review process was confirmed to fall beneath the high standards expected by Materials Today: Proceedings. The veracity of the conference also remains subject to serious doubt and therefore the entire Proceedings has been withdrawn in order to correct the scholarly record.

Big Data Analytics (B.D.A.) is a fast-growing field with the capability to provide useful, clear, and in-depth understanding solutions to healthcare applications. There are many dimensions of big data having issues with its usage, such as managing volume, velocity, variety veracity and value, characteristics such as integrity accuracy and interpretation. However, such challenges do not restrain us from using and exploring big data as a source of evidence in clinical application. This obtains the need to examine health care information to control and reduce the increased cost of medical treatment and improve patient's treatment outcomes. The goal is to describe the ability and capability of big data analytics in health care. This paper describes a review on the developing field of big data analytics in healthcare, discusses the advantages, and provides an architectural framework and methodology. It also investigates the data sources and suggests the platforms/tools for analytics, techniques, technologies, and also presents the challenges and conclusions. This paper concludes that Big data analytics in health care is a promising field for providing more in-depth insight from the massive volume of data sets and functional outcomes.